The Battle for Wesnoth Wiki - User contributions [en]

Vector Inking

2009-01-13T20:07:14Z

Sgt. Groovy:

In art, the term "inking" refers to making a clearly defined lineart from a sketch. The term derives from the ink pens used for this purpose in dead-tree drawing, and the term has transitioned into digital art as well.

Usually the inked lineart is wanted as solid and sharp-edged, and for this vector programs suit well. The added bonus of lineart in vector is that they are freely scalable to any size without loss of resolution. Vector linart can be combined with colouring and shading done in raster by rendering the lineart into raster in desired resolution.

In this tutorial the free vector drawing program [http://www.inkscape.org Inkscape] is used, though many other programs have similar functionality. The tutorial also presumes the version 0.46, the latest stable one, for some specific tools. Basic knowledge of Inkscape's functionality and user interface is expected. Several alternative approaches will be discussed, some of them requiring a tablet, some not.

For all methods you first need the sketch image imported into Inkscape. Put it in its own layer and set the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Filters-Compositing.html#Filters-Blend blend mode] of the layer to "Multiply" and then lock the layer. This way, when you draw everything else on a layer under the sketch layer, the white parts of the sketch appear transparent and the lines are not covered by what you draw. Toggling the visibility of the sketch layer allows you to check how the drawing is coming along.

== Calligraphy ==

The most obvious and intuitive way to ink is by using the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Calligraphy calligraphy tool]. Generally you'll want to set the thinning fixation to 0 so that the line width is determined by the tablet pressure alone. The "mass" property allows you to set the "inertia" of the pen, stabilising the stroke. The calligraphy tool produces paths with lots of nodes, but the node count can usually reduced greatly by simplifying the paths with CTRL-L ("Path -> Simplify") without changing the shape too much (you can set the simplifying treshold in the preferences menu if it simplifies too strongly). As a general rule, you should always simplify paths produced with such tools as calligraphy and tracing, because the lower node count means smaller file and better UI responsiveness, but also makes tweaking by node editing much easier. You can simplify all the strokes at once simply by selecting them all and hitting CTRL-L, and the operation will be done for each path separately. If you intend to fuse the strokes together into one path with "Path -> Union", you should do the simplifying ''first'' to avoid the rounding of the sharp corners in the stroke intersections.

== Bezier tracing with the mouse ==

The problem with calligraphy tool is that in practice it is only useful when used with a tablet (in theory you can change the line width with arrow keys while drawing with mouse, but this is very cumbersome). If you only have a mouse, you can trace the lines with the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Bezier bezier tool], set the stroke width to the desired average for each line, do "[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Stroke-To-Path Path -> Stroke to Path]" and then tweak the line widths by node editing or the new [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Tweak.html Tweak tool].

There are to ways to trace with the mouse, with the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Freehand freehand tool], where you move the cursor along the line and nodes are created on the fly, or the bezier tool, where you set the nodes explicitly. If you have a tablet and a very steady hand, you can manage with the freehand tool, but otherwise the bezier is better. The good thing about the bezier tool is that you end up with relatively few nodes, which keeps the file size smaller and makes later editing of the paths easier. Beginners usually make the beziers by first making an approximating polygon, and then coming back to round the corners. This is most inefficient, a faster and equally easy (after a bit of practice) way is the "tangential sweep." Follow the line and whenever there is a sharp corner, click on it without dragging. In between, click and drag tangentially to the line in few points along the curving line. You learn by practice where to place the sweep points. If there is a straight line between to corners, you don't need to put any nodes there.

Once you have the paths in place and the "Stroke to Path" applied, you must tweak the line widths to give them a more lively look. Before you go for the tweak tool, there is a quick and dirty method worth trying. By selecting all the nodes on one side of the line (or inner/outer subpath of a closed line) and then moving them a bit it's possible to create line width variation for the whole line. The image below illustrates this, on both examples the selected nodes have been moved slightly upwards (to select the nodes on one side of the line, hover the mouse over a node in the middle of the would-be selection, press down Control and use the mouse wheel (or PgUp and PgDown) to increase/decrease the selection. To select all the nodes of a subpath, select one of them and press Ctrl-A). This method often doesn't make the line widths exactly like you want them, but often it's a handy way to do the bulk of the work, completed by little tweaking here and there.

http://koti.welho.com/thonkasa/Roinaa/ink3.png

The first thing you probably want to do with the tweak tool is to make the free ends of lines sharply tapering instead of round or square. The weak tool does this nicely, but it will also shorten the line a bit. You can lengthen the line back by node editing, or you can just anticipate this effect and always put the ends of the lines a bit further that what they are supposed to be when you are doing the tracing.

One thing to know about the tweak tool is that it will operate on ''all'' the nodes of the path, not only the ones under the cursor, and if the path is complex, the tool may become poorly responsive. Simplifying the path will help, as will cutting the subpaths into separate paths (you can keep several paths editable by the tweak tool simultaneously by selecting them all). Also, if you intend to combine several strokes into one path by the Union path operation, you should do all the tweaking first.

== Inkless inking ==

If you have done digital inking before for images that are also coloured, you may have run into the problem that you are forced to trace much of the linework twice — once for the lines and second time for the colour flats. The new [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Bucket.html Paint bucket tool] of the 0.46 version of Inkscape will help a lot, but there is another way to skip the double work. You simply trace the colour flats and leave gaps in between them. With a dark background the gaps will look like lines.

I know what you are thinking: If you have flats with gaps in between, don't you still need to trace the lines twice, once for each side of the gap? Not if you make a silhouette of the whole object first, then trace the flats, cut them off the silhouette with "[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Combining.html Path -> Division]" and in the end shrink all the flats with "[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Editing.html#Paths-Offsets Path -> Inset.]" The procedure is illustrated in the image below. On top left is a sketch of a pitcher. On top middle the outer edge has been traced with path set to dark grey. On top right a copy of the outline path has been created (light gray) and one of the flats has bee traced, filled with red. Notice that the outer edge has been traced already, so the red path needs to siply surround these lines. Both the red and the light grey paths are selected and "Path -> Division" is applied. This will cut the grey path into two pieces, following the red path's edge.

On bottom left the tracing and division has been done to all the flats (the edges of the flats actually align perfectly, the lines you see are rendering artifacts). On bottom middle "Path -> Inset" has been applied to all the flats. The dark lines are the gaps between the flats, showing the dark outline path underneath. These gaps can be tweaked as any paths, but notice that the thinning operation will thicken them and the thickening will thin, since you are operating on the ''surrounding'' paths, not the gaps themselves! In the final picture the gaps have been tweaked to create width variation and to correct some flaws, the flats have been shaded with gradients and a shadow has been made from a copy of the outline path by skewing and blurring.

http://koti.welho.com/thonkasa/Roinaa/ink22.png

== Automatic tracing ==

If you already have a raster version of your linart you can trace it into vector form automatically. If you have [http://potrace.sourceforge.net/ Potrace] installed on your system, you can use them directly from Inkscape ("[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Trace.html Path -> Trace Bitmap]"). Scanned drawings often need some preprocessing in a raster program for a better result, such as increasing contrast, removing spots, smudges and other defects, etc. You can also do the tracing in the raster program — for example in GIMP you can save paths, which are essentially bezier curves, in SVG format. Create the path by creating a selection, with the Magic wand or Select by Color tool, which contains all the pixels you consider to be within the lines, do "Select -> To Path" and after this, in the Paths dialog, choose "Export Path." The result is a SVG file that you can open in Inkscape.

As mentioned in the beginning, having lineart in vector format means that you can scale it up without loss of resolution. In the picture below the small image on top has been rendered from a path. The bottom right image is the same scaled up 300%. Even with cubic interpolation the pixellation of the original image is visible, and the edges have become fuzzy. The bottom left image has been scaled up as a path, and rendered after the scaling. It's edges are every bit as smooth and sharp as the original's.

http://koti.welho.com/thonkasa/Roinaa/ink5.png

Even rerendering in same scale often has an enhancing effect. In the picture below, the top left image has been drawn with a sharp-edged brush without anti-aliasing, resulting in ugly jagginess. The image left is the same, traced into vector and rerendered in same scale. The result is smooth and properly antialiased. The bottom left image has been drawn with a fuzzy brush and the result is also fuzzy. Rerendering has sharpened the image and also cleared some line unstability.

http://koti.welho.com/thonkasa/Roinaa/ink7.png

== Combining vector and raster ==

While vector grapichs are good for lineart, many shading styles, especially more painting-oriented ones, work better in raster. Once you have the lineart in vector, you can render it in raster at the definition you are going to make the shading with and import it into the raster program as a layer. Inkscape even lets you save the files directly in GIMP's XCF format. Everything is rendered into raster and the layer structure of the Inkscape file is transferred into GIMP layers. On the other hand, you can also import SVG files into GIMP as paths and render them there by changing the path into a selection and bucket-filling it. This approach has the benefit that the lineart remains as a path in the GIMP file, and if you decide to scale the image up later, you can use the path to rerender the lineart and thus keep it sharp (the colour layers, not depending on sharp transitions between colours, handle scaling up much better than lineart).

== The stroked skeleton ==

One more interesting and potentially expressive way to do the inking is by first creating a "skeleton" of the linework and then applying a stroking on it by using the "Pattern along Path" [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Effects-GenerateFromPath.html#Effects-PatternAlongPath Effect] or [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-LivePathEffects.html#Paths-LivePathEffects-PatternAlongPath Life Path Effect] (the same tool exist in both menus, with some minor differences). The skeleton consists of a set of paths (or subpaths) each indicating the shape and direction of one brushstroke. A path (or more complex object) is then used as a "brush" and applied to the skeleton paths. In the picture below the top (red) object is the skeleton and the rest are different renderings of it with different brushes (the brush used is next to each version). You can see how the shape of the brush results in very different looks: the middle left looks like it was drawn with a felt-tip pen, the middle right looks like a pencil sketch, bottom left was painted with a solid, wet brush and the bottom right with a coarse, dry one.

http://koti.welho.com/thonkasa/Roinaa/ink1.png

Automated tools like this rarely result in a perfect image, but with a little practice you should able to most of the work with it and complete the image by hand editing. Also, this method lets you easily make different versions of the same image for different purposes. Once you have the skeleton in place trying out different brushes in very easy (if you are using it from the LPE menu, any changes in the brush are updated in the rendered image immediately!) From the above image you can see that the strokes of different length can have a very different look, due to the stretching. One step to improve the method would be to divide the subpaths on the skeleton into groups by length (and possibly by role, that is whether they are part of the outlines or the shading) and use different brushes on them.

== More automated methods ==

If your linework is supposed to include aspects of shading, made by hatching or stippling, there are further tools to create them automatically by applying multiple copies of patterns or brushes in a controlled manner. Complex hatching can be created by the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-LivePathEffects.html#Paths-LivePathEffects-StitchSubPaths Stitch subpaths] LPE or the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Effects-GenerateFromPath.html#Effects-Interpolate Interpolate Effect], further edited by the tweak tool. As a new feature in the 0.46 version, the calligraphy tool can be used with a [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Calligraphy-Hatchings guide path], meaking drawing parallel lines much easier. Also, the trace background feature can be use to make the hatching line thickness follow a preset shading pattern. For stippling, the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Tiles.html Tile clones tool] with its [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Spheres-Dots.html trace background feature] may be the most useful.

[[Category:Art Tutorials]]

Create Art

2008-06-16T08:41:51Z

Sgt. Groovy: new tutorial added

{| style="float:right; margin-left:1em;"
|__TOC__
|}

== Getting Started ==

=== General Information ===

Graphic artists usually meet on the [http://www.wesnoth.org/forum/viewforum.php?f=9 artwork development forum] or on the [http://www.wesnoth.org/forum/viewforum.php?f=18 restricted art development forum]. The former is a great place to post and discuss new and current Wesnoth art and graphics, and the latter to see what the art development team is working on.

Unit and terrain art is stored in the portable network graphics (PNG) format. Each frame of a unit animation, and each variation of a terrain is stored as a separate .png file in the "images" subdirectory of wesnoth, and generally these files will be 72 x 72 pixels (the size of Wesnoth's basic hexagonal tile) with an alpha channel (a part of the file that indicates how transparent each pixel is). When creating your own images, you can test them without overwriting any game data by putting them in your userdata directory (see [[EditingWesnoth]] for details on Wesnoth's directory structure). The game also supports JPEG images, though these are better suited for story art.

To edit these graphics, you'll need some program capable of creating PNGs - some of the programs in the following list are free, open-source software, and will do the job nicely: [[Art Programs]]

If you need some inspiration, we have a [[GraphicLibrary|Graphics Library]] which collects art posted on the forum. You can use this for ideas, and as a scrap heap for different parts of unit images (a technique described [[Give Your Hero A Personality|here]]).

=== Roadmap ===
A list of what's being done and what needs doing:
* '''[[Tiles Status]]''' - a roadmap/plan of sorts for future work on terrain tiles
* [http://www.wesnoth.org/forum/viewtopic.php?t=9162&sid=21cf168d2b077d06bb0017c059b4add6 A list] of current work that needs to be done with sprites.

== Art Tutorials ==

These are a work-in-progress, and describe both how to make art fit into wesnoth's style, as well as giving some considerable tips on drawing in general. Especially useful is the [[External Tutorials]] page which lists a large number of art tutorials available on the web.

=== General Art and Computer Graphics ===
* [[Using the Levels Adjustment]] - making scanned pencil drawings ''not'' look washed out.
* [[Extending dynamic range]] - The Grooviest (so far) tutorial about extending the dynamic range of images and how this technique can be used to make better scans of pencil drawings.
* [[Scanning with camera]] - How to transfer real-life art to computer using a digital camera instead of a scanner.
* [[How to Anti-Alias Sprite Art]] - A means of removing the jagged edges on pixel lines
* [[Art Supplies]] - What physical items you need to do larger cell-shaded art like that of Jetryl/Jormungadr/et al
* [[Inking With Pencils|Computer Inking a Sketch]] - Info from Jason Lutes on his portrait workflow
* [[Scaling Digital Images]] - how to properly resize an image on a computer
* [[How to Shade]] - at attempt at tackling a very complicated topic
* [[Cartography for Wesnoth|How to make Wesnoth-style Maps ]] - Kestenvarn's tricks of the trade
* [[Designing weapons and armour]] - Advice from zookeeper on designing realistic weapons for your characters
* [[Portrait Tutorial]] - a guide on how to draw unit/character portraits by Kitty.
* Sgt. Groovy's vector workshop - tips and tricks for drawing with Inkscape
** [[Z-order tricks]] - few methods for faking overlapping shapes
** [[Variable-width strokes]] - how to make the strokes vary in width, like being drawn with a flat-tipped pen — no tablet needed!
** [[Shaped gradients with Gaussian blur]] - how to make gradients in other shapes than linear or radial
** [[Smooth shading in vector]] - the basic vector techniques for smooth shading, employing Gaussian blur and clipping/masking
** [[Vector Inking]] - vector techniques, including mouse-only, for inking your sketches
** [[Making portrait art in vector]] - a complete tutorial for making Wesnoth unit portraits in vector graphics.

=== [[Terrain|Terrain Graphics]] ===

The following is information specific to drawing terrain for Wesnoth. Read Frame's "Tiles Tutorial" for a good overview of how terrain graphics work.

* [[Tiles Tutorial]] - Frame's tutorial describing the process of making terrain tiles in wesnoth, and how they interact with adjacent tiles.
* [[How To Make Seamless Tiles]] - The tutorial is aimed at Photoshop users, but the technique is similar with The GIMP.
* [[Seamless Tiles Using Inkscape]] - This tutorial teaches a method for making seamless hex tiles in vector craphics (to be rendered in raster).
* [[Turning Square Tiles into Hex]] - Nifty tricks for transforming square (or any rectangle) shaped seamless tiles into hexagon seamless tiles.
* [[CastleTutorial|Castle Tutorial]] - A description of how Wesnoth's castle tiles work (needs updating, but useful nonetheless)
* [[MultiHexTutorial|Multi-Hex Tiling Tutorial]] - A description of how multi-hex tiles work.
* [[Editing Castles]] - Instructions for how to make/edit castles (and other corner-based terrains) using yobbo's GIMP script.

These describe the system used to specify how terrains behave in game:
* [[TerrainCodesWML]] - A list of the letters used to represent terrain types.
* [[TerrainGraphicsWML]] - If you really need to get technical, start here.
* [http://www.anathas.org/ayin/wesnoth/doc/terrain_graphics_wml Ayin's Terrain Graphics document] - If you really, ''really'' need to get technical, this describes the terrain graphics WML system in depth.

=== Sprite Art ===
The following are different tutorials about sprite work compiled by various wesnoth sprite artists. These will give you the most specific-to-wesnoth information about making sprites, and are well worth a read.

* [[Creating Unit Art]] - a list of specifications you will need to match.
* [[Give Your Hero A Personality]] - tricks for editing existing images, including some clip art.
* [[Basic Animation Tutorial]] - or "How to Animate Sprites for Dummies," covering the basic theory, and all of the mistakes to avoid.
* [[Advanced Animation Tutorial]] - how to wire in the animation frames to units in a way that makes the animation look good in-game. Doesn't cover any drawing as such, but focuses on timing conventions and some standard techniques for getting as much as possible out of the frames that are already drawn.
* [[Team Color Shifting]] - how to use our new team color system.
* [[Creating Shadows Under Units]] - how we create the shadows for the units in-game
* [[Making Bow Animations]] - the current standard for how we want bow animations to work
* [[Rotate Pixel Art Without Blurring]]
* [[Creating a scratch built sprite]] - an attempt to show some ways creating a sprite from scratch
* [[How to create motion blurs]] - a simple explanation on how to create attack animation weapon blurs
* [[FrankenPacks]] - a quick and dirty way to create sprites for [[WesnothAcronyms|UMC]]

=== External Tutorials ===
The following page contains dozens of links to tutorials covering all manner of artwork, including sprite art. These were not made by wesnoth artists, but should prove very useful for general instruction.

* [[External Tutorials]]

== See Also ==
* [[Create]]
* [[EditingWesnoth]]
* [[GraphicLibrary]] - Lots of usercontributed graphics (most should be GPL'ed)
* [[Project]]
* [http://wesnoth.dbzer0.com/blog/wpg2 External Graphic Library] - A project to better organize the art of Wesnoth.

[[Category:Art Tutorials]]

Vector Inking

2008-06-16T08:40:54Z

Sgt. Groovy: New tutorial added

In art, the term "inking" refers to making a clearly defined lineart from a sketch. The term derives from the ink pens used for this purpose in dead-tree drawing, and the term has transitioned into digital art as well.

Usually the inked lineart is wanted as solid and sharp-edged, and for this vector programs suit well. The added bonus of lineart in vector is that they are freely scalable to any size without loss of resolution. Vector linart can be combined with colouring and shading done in raster by rendering the lineart into raster in desired resolution.

In this tutorial the free vector drawing program [http://www.inkscape.org Inkscape] is used, though many other programs have similar functionality. The tutorial also presumes the version 0.46, the latest stable one, for some specific tools. Basic knowledge of Inkscape's functionality and user interface is expected. Several alternative approaches will be discussed, some of them requiring a tablet, some not.

For all methods you first need the sketch image imported into Inkscape. Put it in its own layer and set the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Filters-Compositing.html#Filters-Blend blend mode] of the layer to "Multiply" and then lock the layer. This way, when you draw everything else on a layer under the sketch layer, the white parts of the sketch appear transparent and the lines are not covered by what you draw. Toggling the visibility of the sketch layer allows you to check how the drawing is coming along.

== Calligraphy ==

The most obvious and intuitive way to ink is by using the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Calligraphy calligraphy tool]. Generally you'll want to set the thinning fixation to 0 so that the line width is determined by the tablet pressure alone. The "mass" property allows you to set the "inertia" of the pen, stabilising the stroke. The calligraphy tool produces paths with lots of nodes, but the node count can usually reduced greatly by simplifying the paths with CTRL-L ("Path -> Simplify") without changing the shape too much (you can set the simplifying treshold in the preferences menu if it simplifies too strongly). As a general rule, you should always simplify paths produced with such tools as calligraphy and tracing, because the lower node count means smaller file and better UI responsiveness, but also makes tweaking by node editing much easier. You can simplify all the strokes at once simply by selecting them all and hitting CTRL-L, and the operation will be done for each path separately. If you intend to fuse the strokes together into one path with "Path -> Union", you should do the simplifying ''first'' to avoid the rounding of the sharp corners in the stroke intersections.

== Bezier tracing with the mouse ==

The problem with calligraphy tool is that in practice it is only useful when used with a tablet (in theory you can change the line width with arrow keys while drawing with mouse, but this is very cumbersome). If you only have a mouse, you can trace the lines with the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Bezier bezier tool], set the stroke width to the desired average for each line, do "[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Stroke-To-Path Path -> Stroke to Path]" and then tweak the line widths by node editing or the new [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Tweak.html Tweak tool].

There are to ways to trace with the mouse, with the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Freehand freehand tool], where you move the cursor along the line and nodes are created on the fly, or the bezier tool, where you set the nodes explicitly. If you have a tablet and a very steady hand, you can manage with the freehand tool, but otherwise the bezier is better. The good thing about the bezier tool is that you end up with relatively few nodes, which keeps the file size smaller and makes later editing of the paths easier. Beginners usually make the beziers by first making an approximating polygon, and then coming back to round the corners. This is most inefficient, a faster and equally easy (after a bit of practice) way is the "tangential sweep." Follow the line and whenever there is a sharp corner, click on it without dragging. In between, click and drag tangentially to the line in few points along the curving line. You learn by practice where to place the sweep points. If there is a straight line between to corners, you don't need to put any nodes there.

Once you have the paths in place and the "Stroke to Path" applied, you must tweak the line widths to give them a more lively look. Before you go for the tweak tool, there is a quick and dirty method worth trying. By selecting all the nodes on one side of the line (or inner/outer subpath of a closed line) and then moving them a bit it's possible to create line width variation for the whole line. The image below illustrates this, on both examples the selected nodes have been moved slightly upwards (to select the nodes on one side of the line, hover the mouse over a node in the middle of the would-be selection, press down Control and use the mouse wheel (or PgUp and PgDown) to increase/decrease the selection. To select all the nodes of a subpath, select one of them and press Ctrl-A). This method often doesn't make the line widths exactly like you want them, but often it's a handy way to do the bulk of the work, completed by little tweaking here and there.

http://user.personal.fi/atk/phpopintopiiri/kuvia/ink3.png

The first thing you probably want to do with the tweak tool is to make the free ends of lines sharply tapering instead of round or square. The weak tool does this nicely, but it will also shorten the line a bit. You can lengthen the line back by node editing, or you can just anticipate this effect and always put the ends of the lines a bit further that what they are supposed to be when you are doing the tracing.

One thing to know about the tweak tool is that it will operate on ''all'' the nodes of the path, not only the ones under the cursor, and if the path is complex, the tool may become poorly responsive. Simplifying the path will help, as will cutting the subpaths into separate paths (you can keep several paths editable by the tweak tool simultaneously by selecting them all). Also, if you intend to combine several strokes into one path by the Union path operation, you should do all the tweaking first.

== Inkless inking ==

If you have done digital inking before for images that are also coloured, you may have run into the problem that you are forced to trace much of the linework twice — once for the lines and second time for the colour flats. The new [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Bucket.html Paint bucket tool] of the 0.46 version of Inkscape will help a lot, but there is another way to skip the double work. You simply trace the colour flats and leave gaps in between them. With a dark background the gaps will look like lines.

I know what you are thinking: If you have flats with gaps in between, don't you still need to trace the lines twice, once for each side of the gap? Not if you make a silhouette of the whole object first, then trace the flats, cut them off the silhouette with "[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Combining.html Path -> Division]" and in the end shrink all the flats with "[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Editing.html#Paths-Offsets Path -> Inset.]" The procedure is illustrated in the image below. On top left is a sketch of a pitcher. On top middle the outer edge has been traced with path set to dark grey. On top right a copy of the outline path has been created (light gray) and one of the flats has bee traced, filled with red. Notice that the outer edge has been traced already, so the red path needs to siply surround these lines. Both the red and the light grey paths are selected and "Path -> Division" is applied. This will cut the grey path into two pieces, following the red path's edge.

On bottom left the tracing and division has been done to all the flats (the edges of the flats actually align perfectly, the lines you see are rendering artifacts). On bottom middle "Path -> Inset" has been applied to all the flats. The dark lines are the gaps between the flats, showing the dark outline path underneath. These gaps can be tweaked as any paths, but notice that the thinning operation will thicken them and the thickening will thin, since you are operating on the ''surrounding'' paths, not the gaps themselves! In the final picture the gaps have been tweaked to create width variation and to correct some flaws, the flats have been shaded with gradients and a shadow has been made from a copy of the outline path by skewing and blurring.

http://user.personal.fi/atk/phpopintopiiri/kuvia/ink22.png

== Automatic tracing ==

If you already have a raster version of your linart you can trace it into vector form automatically. If you have [http://potrace.sourceforge.net/ Potrace] installed on your system, you can use them directly from Inkscape ("[http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Trace.html Path -> Trace Bitmap]"). Scanned drawings often need some preprocessing in a raster program for a better result, such as increasing contrast, removing spots, smudges and other defects, etc. You can also do the tracing in the raster program — for example in GIMP you can save paths, which are essentially bezier curves, in SVG format. Create the path by creating a selection, with the Magic wand or Select by Color tool, which contains all the pixels you consider to be within the lines, do "Select -> To Path" and after this, in the Paths dialog, choose "Export Path." The result is a SVG file that you can open in Inkscape.

As mentioned in the beginning, having lineart in vector format means that you can scale it up without loss of resolution. In the picture below the small image on top has been rendered from a path. The bottom right image is the same scaled up 300%. Even with cubic interpolation the pixellation of the original image is visible, and the edges have become fuzzy. The bottom left image has been scaled up as a path, and rendered after the scaling. It's edges are every bit as smooth and sharp as the original's.

http://user.personal.fi/atk/phpopintopiiri/kuvia/ink5.png

Even rerendering in same scale often has an enhancing effect. In the picture below, the top left image has been drawn with a sharp-edged brush without anti-aliasing, resulting in ugly jagginess. The image left is the same, traced into vector and rerendered in same scale. The result is smooth and properly antialiased. The bottom left image has been drawn with a fuzzy brush and the result is also fuzzy. Rerendering has sharpened the image and also cleared some line unstability.

http://user.personal.fi/atk/phpopintopiiri/kuvia/ink7.png

== Combining vector and raster ==

While vector grapichs are good for lineart, many shading styles, especially more painting-oriented ones, work better in raster. Once you have the lineart in vector, you can render it in raster at the definition you are going to make the shading with and import it into the raster program as a layer. Inkscape even lets you save the files directly in GIMP's XCF format. Everything is rendered into raster and the layer structure of the Inkscape file is transferred into GIMP layers. On the other hand, you can also import SVG files into GIMP as paths and render them there by changing the path into a selection and bucket-filling it. This approach has the benefit that the lineart remains as a path in the GIMP file, and if you decide to scale the image up later, you can use the path to rerender the lineart and thus keep it sharp (the colour layers, not depending on sharp transitions between colours, handle scaling up much better than lineart).

== The stroked skeleton ==

One more interesting and potentially expressive way to do the inking is by first creating a "skeleton" of the linework and then applying a stroking on it by using the "Pattern along Path" [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Effects-GenerateFromPath.html#Effects-PatternAlongPath Effect] or [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-LivePathEffects.html#Paths-LivePathEffects-PatternAlongPath Life Path Effect] (the same tool exist in both menus, with some minor differences). The skeleton consists of a set of paths (or subpaths) each indicating the shape and direction of one brushstroke. A path (or more complex object) is then used as a "brush" and applied to the skeleton paths. In the picture below the top (red) object is the skeleton and the rest are different renderings of it with different brushes (the brush used is next to each version). You can see how the shape of the brush results in very different looks: the middle left looks like it was drawn with a felt-tip pen, the middle right looks like a pencil sketch, bottom left was painted with a solid, wet brush and the bottom right with a coarse, dry one.

http://user.personal.fi/atk/phpopintopiiri/kuvia/ink1.png

Automated tools like this rarely result in a perfect image, but with a little practice you should able to most of the work with it and complete the image by hand editing. Also, this method lets you easily make different versions of the same image for different purposes. Once you have the skeleton in place trying out different brushes in very easy (if you are using it from the LPE menu, any changes in the brush are updated in the rendered image immediately!) From the above image you can see that the strokes of different length can have a very different look, due to the stretching. One step to improve the method would be to divide the subpaths on the skeleton into groups by length (and possibly by role, that is whether they are part of the outlines or the shading) and use different brushes on them.

== More automated methods ==

If your linework is supposed to include aspects of shading, made by hatching or stippling, there are further tools to create them automatically by applying multiple copies of patterns or brushes in a controlled manner. Complex hatching can be created by the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-LivePathEffects.html#Paths-LivePathEffects-StitchSubPaths Stitch subpaths] LPE or the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Effects-GenerateFromPath.html#Effects-Interpolate Interpolate Effect], further edited by the tweak tool. As a new feature in the 0.46 version, the calligraphy tool can be used with a [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Paths-Creating.html#Paths-Calligraphy-Hatchings guide path], meaking drawing parallel lines much easier. Also, the trace background feature can be use to make the hatching line thickness follow a preset shading pattern. For stippling, the [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Tiles.html Tile clones tool] with its [http://tavmjong.free.fr/INKSCAPE/MANUAL_v14Draft/html/Spheres-Dots.html trace background feature] may be the most useful.

Making portrait art in vector

2008-05-28T13:21:58Z

Sgt. Groovy:

''NOTE: The style discribed here is no longer used in Wesnoth mainline portraits.''

This is a tutorial for making unit portraits for the Battle for Wesnoth in vector graphics. The Wesnoth portraits are made in what is called "the Lutesian style" after the comic artist Jason Lutes, who has donated many of the Loyalist portraits. The main reason for stylistic restrictions is to achieve aesthetic consistency.

The technical aspects of the Lutesian style can be summarised as follows:

* The colouring is made in "cell shading," that is, colours are applied in constant colour areas, known as "flats." Changes in tone (such as shadows and highlights) are also made the same way, instead of brush strokes or gradients.

* All the linework, instead of being black or having another colour of their own, is made in darker (or sometimes lighter) shade of the colour of the flat it is on top of.

* Adjacent flats don't need to be separated by outlines.

In addition to this, the final format of the portrait file should be a 205x205 pixel PNG with a fully black background.

Vector graphics programs can be easily used to produced this kind of images, because the flats can be easily crated and edited as paths filled with a solid colour. The linework, shadows and outlines are most conveniently made as semitransparent dark or light areas on top of the flats, because then they always follow the colour of the flats, if it is changed.

In this tutorial we use Inkscape as the program, but any vector drawing program worth its salt should have the same functionality. As a prequisite, you should be familiar with the basic concepts of vector drawing, how to make, edit and stroke paths, the fill rules (how to make transparent "holes" in paths), etc. Basically, reading the Inkscape manual and playing around with it for half an hour should be enough. You should also be able to draw and scan a pencil sketch four your portrait, or have other people's sketches available (remember to ask for permission).

Here's our example sketch, a young burgher:

http://koti.welho.com/thonkasa/Roinaa/burghersketch2.png

The blue lines show the edges of the shadows. It is convenient to make the shadow edges visible, because it is then easier to trace them (and scanners easily lose the lighter shades of pencilwork anyway), and by using a colour pencil, you make them separable from the actual linework. You can do the same thing with highlights, but use different colour than for shadows. The red arrow indicates the general direction of lighting.

The first thing to do is to trace out the flats and make outlines for those that are going to have them. As stated above, we are going to make the outlines as a separate, semitransparent path on top of the flat path, so that it gets its colour from it. Because lines differentiating two flats of different colours should have the colour of one or the other, but not both, it follows that the outer edge of the outline path must align exactly with the edge of the flat path. We will learn shortly how to accomplish this, but first read [[Z-order tricks|this tutorial]] about the z-order and the path tools.

Then it's time to get work on the image. First thing to do is to prepare the sketch. Since we must keep the sketch as the topmost object to see it as we draw, but also see the work beneath it, the background colour of the skecth must be changed into transparency. This can be done in GIMP, where the necessary tool can be found in "Layers -> Transparency -> Color to alpha." Once this is done, save it as PNG and import it into Inkscape. It will initially have some random X and Y values, so change them to 0, in the top toolbar. Open the document preferences and set the page size to match the selection. Rename the current layer to "sketch" and lock it, so that you don't accidentally move the sketch image, and it will be easier to select the objects under it. Create a new layer under the sketch layer and name it "flat." Now you are ready to trace the flats.

There are to ways to trace, with the freehand tool, where you move the cursor along the line and nodes are created on the fly, or the bezier tool, where you set the nodes explicitly. If you have a tablet and a very steady hand, you can manage with the freehand tool, but otherwise the bezier is better. The good thing about the bezier tool is that you end up with relatively few nodes, which keeps the file size smaller and makes later editing of the paths easier. Beginners usually make the beziers by first making an approximating polygon, and then coming back to round the corners. This is too inefficient, a faster and equally easy (after a bit of practice) way is the "tangential sweep." Follow the line and whenever there is a sharp corner, click on it without dragging. In between, click and drag tangentially to the line in few points along the curving line. You learn by practice where to place the sweep points. If there is a straight line between to corners, you don't need to put any nodes there.

The flats are best traced in back to front order, so that flats that are going to be the on the bottom are drawn first. It's usually good to base the z-order of the flats to the natural order of the real-life objects they represent, so that flats representing the foreground objects are on top of the flats representing background objects. There are times when you must break this order, but following it as a general rule helps to keep the drawing organised in your head.

In our image, we start with the sleeve at left. It's one of the "bottom objects," it's being covered by the torso, but doesn't itself cover anything. In the image below we see the sleeve traced with the bezier tool. Note that this is not the only, and not probably even the best, way to place the nodes. There actually isn't a "right" way to trace a shape, the only places where you ''must'' put a node are the sharp corners. You should try to get away with as few nodes as possible, though. You will also notice tha we didn't trace the line separating the sleeve and the torso. We will do it when we trace the flat for the torso, which comes on top of this one.

http://koti.welho.com/thonkasa/Roinaa/burgshot1.png

If we want the flat to have an outline, this is the time to do it, because it will then be directly on top of the flat it follows, and has the same relative z-placement as the flat.

The outline could be made with the calligraphy tool, but then we would be tracing the same lines twice. The flat could also be stroked, which would produce the outline automatically, but there are two problems. First, the stroke is always half outside and half inside the path, but we want the outer edge of our outline to follow the path exactly. Second, the stroke is always constant width, and this tends to result in dull and formalistic look, better suited to diagrams and such than artistic drawings. Even though strokes are not good as themselves, there is a handy tool "stroke to path," which makes a path out of a stroke so that the path follows the edges of the stroke, and we can use this to turn a stroke into a proper outline.

The "stroke to path" tool makes the original path disappear, so we will first do "Edit -> Copy" and "Edit -> Paste in place" for our flat. Then select the copy and disable fill and enable stroke. Set the stroke colour to black with 50% opacity (you might want to adjust this later, but these are good working settings at this point).

http://koti.welho.com/thonkasa/Roinaa/burgshot2.png

The width of the stroke should be about twice the intended mid-width of the outline. Then do "Path -> Stroke to path." Half of the outline still lies outside the path, and we will cut it away with a clip path, but wirst we will have to make the outline variable width. This is done easily by moving the outline path slightly, only so much that the edges of the underlying flat stay within it. Moving the outline makes the part that is inside the flat, that will remain after the intersection operation, to become thinner in some parts and thicker in other. As a general rule, the outline should be thinner on the lighted side of the object and thicker on the shadow side, so you should move the outline towards the light source.

http://koti.welho.com/thonkasa/Roinaa/burgshot3.png

When the inside part of the outline is the way you like, select the flat path and do "Edit -> Clone -> Clone". Select the clone and the outline path and do "Object -> Clip -> Set". What we did was using the clone of the flat as a clip path on the outline path, making the parts that are outside the path invisible. Because we use a clone, if we edit the nodes of the flat, the changes will automatically change the (visible) shape of the outline path correspondingly.

http://koti.welho.com/thonkasa/Roinaa/burgshot4.png

Since this method decides the outline width only by the angle of the outline, it doesn't always produce the desired outcome in all the places, but you can easily adjust the line width by moving the appropriate nodes of the ''inner edge'' of the outline path, inwards to thicken the line, outwards to thin it (the "sclupting" mode can become handy for this).

Proceed by this method for the flats in back-to-front order. Having the flats traced and outlined, my picture would look like this:

http://koti.welho.com/thonkasa/Roinaa/burgher1.png

Note that not all flats need an outline. Outlining usually adds a sense of weight and substance to objects, so thin, light and fluffy objects, such as hair, beard, moustache and the quill in my image, generally look better without.

Later it will be useful to have the silhouette of the whole figure in one path. Create a new layer titled "silhouette," select all the flat paths and do "Edit -> Copy." Switch to the "silhouette" layer and do "Edit -> Paste in place," followed by "Path -> Union." This will fuse all the paths into one, creating a silhouette. Turn the visibility of this layer off and lock it.

At this point it might be a good time to handle some special features, such as eyes. We could make them as flats on top of the face flat, but the pupils usually require adjustements to make them look just right, it's more convenient to make the eyes as holes in the face, and put the pupils behind it. If you make the pupils as whole ovals, only part of them is visible, but when you move them, the part always shows which should.

There are two ways to make a hole in a path. One is to make a path the shape of the hole on top of it and doing "Path -> Difference," but if you select a path and press shift while starting to make a new one, the new one will become a subpath to the selected one, and if it's inside it, it will appear as a hole (provided the fill rule for the path is even-odd, the default).

If you want the holes to have outlines as well, it will require a bit of trickery. Select the face flat and do "Edit -> Copy" and "Edit -> Paste in place." Then do "Path -> Break apart," which will make each of the subpaths into separate paths. Select the outermost path and delete it. What you are left with are paths that excactly cover the holes. Follow the outline making procedure, except in the end, instead of "Path -> Intersection," do "Path -> Difference" to clip away the ''inner'' half of the outline path.

For the eyes, you need white for the eyeballs (the white you see is the background colour). So make a eyeball colour flat (not usually pure white) that covers both of the eye holes, and drop it below the face flat. Then use the ellipse tool to make the iris and the pupil, group them, and drop the group below the face flat, but above the eyeball flat. When you have made one iris/pupil combo, you can copy it to get the other be exactly alike.

http://koti.welho.com/thonkasa/Roinaa/burgshot11.png

Now would be a good time to make the rest of the linework, that is the lines not outlining any flats. These lines are also to be made dark with partial transparency, so that they become a darker shade of the colour of the flat they are covering. There are several ways to make the linework. With short lines extending from the outlines, like the ones on the sleeves in my picture, can be made by making new nodes in appropriate places on the outline path and dragging them out. Thick lines can me treated as the flats by tracing the edges of the lines with the bezier tool. If you have a tablet, the calligraphy tool might be useful, but it takes some practice to use well. If the calligraphy lines overlap, you have to fuse them together by doing "Path -> Union" on them, or the partial transparency will produce darker shades on the overlapping parts.

Sometimes tracing linework is done more easily on a raster program, because you can use the eraser tool to trim the lines and correct mistakes immediately (especially with a tablet, when you can use the other end of the stylus as the eraser). The prosedure is as follows: Open the original sketch bitmap on a raster program. Create a new transparent layer on top and do the linework on this with black. Save only this layer as a PNG and import it into Inkscape. Set the X and Y properties of the bitmap to zeroes to make it align with the rest of the image. Select the bitmap and do "Path -> Trace bitmap" (the default options work well most of the time). Delete the bitmap and what you are left with is the linework done in paths. They are fully black so decrease the opacity and edit them further with the usual tools, if necessary.

Whatever method you use to make the linework, it's good to keep things organised and put it on its own layer.

http://koti.welho.com/thonkasa/Roinaa/burgher2.png

After the linework we can do the shadows. Create a new layer titled "shadow" on top of the others, and start tracing the shadow edges like you did with flats, but obviously, the shadows don't need outlines. You can make the shadow paths extend beyond the outlines of the figure, and cut them off in the end using the silhouette path you created earlier, using "Path -> Intersection." The shadows will also be partially transparent to show through the colour of the flat, but their own colour can well be something else than black. In my image I have used dark blue for the shadows, as in outdoors the light coming to the shadow parts comes from the sky and is bluer than the sunlight. You can try different colours for your shadows to see how it affects the general mood if the image.

http://koti.welho.com/thonkasa/Roinaa/burgher3.png

We still have some details to do, like the buttons on the jacket. Here I made the buttons with the ellipse too, using the same ellipse to create the shadow (by now you should be able to figure out how to make the crescent shape). If you need identical objects, you can always make copies of one, but if you make clones instead, and decide to change something about the design of the object later, you only need to change the original and the clones change as well.

http://koti.welho.com/thonkasa/Roinaa/burgshot12.png

To complete the image, some final touches are needed, like the specular spots in the eyes and some texture and highlights for the beard and hair (all done with semitransparent dark and light paths). Then the standard black background and some cropping and we're all done (I'll leave it as the exercise of the reader to figure out how to do the cropping).

http://koti.welho.com/thonkasa/Roinaa/burgherfinal1.png

[[Category: Art Tutorials]]

Smooth shading in vector

2008-03-30T09:10:48Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

In this tutorial we are going to learn few methods for making smooth shading in vector graphics. Shading is very important in bringing out the three-dimensional shape of objects, and combining Gaussian blur with clipping and masking brings much more power to accomplish it than mere gradients.

The important part with blurring to keep in mind is that the colour transition function is different than in gradients. Gradients have a linear transition (between any two stops), while in blurred objects it is S-shaped. What the difference means in practice is that even if a gradient goes all the way down to full transparency/background colour, it will still produce a distinct egde (top left in the image below). The blurred object, on the other hand, will have truly fuzzy edge (top right). It also makes a difference whether an object that is both blurred and transformed is first blurred or transformed. Bottom left is a square that has first been blurred and then scaled vertically down 50%. Bottom left is the same square, but it is first scaled and then blurred.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial7.png

The order is different when an object is both blurred and clipped/masked. In that case the blurring is always done first, and for that reason you can use clipping/masking to create objects with some edges sharp and some blurred. If you want the object to be blurred after clipping/masking, you have to put it in a group, and clip/mask the group (in Inkscape, you can't group single object, so you have to create a dummy object, group it with the real object, CRTL-select the dummy and remove it).

Let's start with a very simple example, a sphere. You probably already know how to do a shaded sphere with radial gradient, so here's another kind of method: First make a circle and fill it with the colour of the shadow side (1). Make a copy of it in the lighted side colour (lighter and more saturated version of the shadow colour), move it toward the light source (here up and to the left) and blur it (2). Make another copy of the original circle and use it as a clip path on the blurred circle. You'll see that the clipping cut off all the "leaking" blurriness, and the rest will make a nice smooth shading on the sphere. Let's make a shadow for the sphere too, while we're at it: First make a circle of the shadow colour and blur it (4) (it's most often a good idea to make the shadow as it looks from the above, and use transformations to bring it into right perspective). Real live shadows have the quirky property that their blurriness depends on the distance from the object casting them. The blur radius of gaussian blur filter is the same everywhere, but we can fake the effect with a mask. Make another circle, somewhat bigger than the shadow, white in fill colour and blur it, but less than you blurred the shadow (5). Use this as a mask on the shadow and you get nice variably blurred shape (6). With little scaling and rotating it looks good to go under the sphere (7).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial5.png

For a sphere you could have made a decent shading with radial gradient as well, but this approach works with other kind of shapes too. Sometimes you can also use the blurred path to make the shadow side, like in the image below.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial6.png

You can also use several blurred paths for shading, like below. The middle lightness (clipped) path was made by copying the original path, and the lightest one by making a linked offset (left). When both are blurred (with different blur radius) they result in a bit glossier surface than in the sphere example (middle). The technique works pretty well even for such odd shape, but it's not perfect. That's OK, there's nothing stopping you from editing the blurred paths, like has been done with the rightmost image (you will have to do "Path -> Object to Path" for the linked offset in order to node edit it). If you use a clone of the base path for clipping, you can still easily edit the base path after shading, because the clip path updates automatically.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial11.png

The above approach works best with smooth, organig shapes, but blurring can come handy when doing edgy shapes with flat shading. In the image below the edges are made smoother and more real life-like by a bit of blurring.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial8.png

Usually clipping/masking is uset to get rid of the "leaking" blurriness outside the path, but the "leaking" can also be used to one's advantage. In the image below there is a circle with a bigger one around it, a hole being cut in it's middle the size of the smaller circle. The bigger circle has been filled with a gradient (1). When the bigger circle is blurred, the blur will leak over the smaller one (2). When a copy of the small circle is used as a clip path on the bigger one, only the leak is left visible, creating a shading on the small circle (3). When a bigger blur radius is used, the shape changes from flat, button type one, to rounder one (4).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial9.png

Here's one more variation of these techniques, to produce a flat, paint splash type shape. First make the base path (1). Make a copy of it in darker, less saturated colour and move toward the shadow side (1). use a copy of the base bath to cut off ("Path -> Difference") parts of the shadow path and move back a bit (3). Blur the shadow path (4). Following the same procedure, create the highlight path (lighter more saturated, on the opposite side) (5). Clip both the blurred paths with a clone of the base path (6).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial10.png

==External links==

Here's some links to similar tutorials:
* More [http://howto.nicubunu.ro/inkscape_blur_orbs_3d_text/ blur shading] from Nicu (her other tutorials are also worth checking out)
* Awesome [http://www.inkscape.org/screenshots/gallery/inkscape-0.45-3D-rope.png "interactive" 3-d rope], demonstrating the power of clones
* Fairly complex [http://troy-sobotka.blogspot.com/2007/11/inkscape-tutorial-1-chrome-effect.html "chrome" text effect], where the text can be changed and the effect updates
* [http://chrisdesign.wordpress.com/2008/02/14/simple-metal-orb-using-gradients/ Realistic metal shading] made with multi-stop gradients

[[Category: Art Tutorials]]

Shaped gradients with Gaussian blur

2008-03-30T08:58:52Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

The gradient tool in Inkscape is incredibly useful, but it has its limitations as well. For example, the radial gradient is limited to circular or oval in shape. To make gradients of other shapes, the Gaussian blur tool becomes handy. When you blur any object different in colour than its background, you are essentially creating a shaped gradient between those colours.

To make a "radial" gradient of any shape, that is, a gradient that changes from the edge of the path toward the centre, first create the path and set its fill colour to what you want the outer colour to be. Then make "Path -> Linked Offset", raise it to the top and set it to the inner colour (or usually somewhat more saturated one, to compensate the mixing when blurred). Use the node handle to make it smaller than the original path (1) and blur it to create the gradient (2). Because the inner path is linked to the outer one, any changes to the outer path will be updated in the shape of the inner one (though you may need to readjust the blur radius if you change the shape radically). To make multi-stop gradient, simply make several linked offsets of different colours and sizes (3). You will notice that the colour transition is smoother (rounder) than in normal gradients. If you need more linear gradient, you need to make several intermediary (in size and colour) linked offset paths.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial2.png

The same method is good for making any odd-shaped gradients. Just create a stepwise gradient with several paths and blur them to make the gradient smooth. In the image below there is a stepwise gradient created with the interpolate effect (this effect is especially handy for this, because it can interpolate the colour as well). In the lower image all but the outermost path have been blurred.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial3.png

One of the limitations of the gradient tool, when applied to strokes, is that if you want the gradient to change from the inner edge to the outer, you can only do this for oval-shaped paths. The blur, together with masking, can fix this problem as well. First create the base path, then make a linked offset path for it, raise it to the top and set its fill to none and the stroke to the outer colour (1). Create another linked offset, raise it to the top and set its fill to the inner colour, stroke to none and blur it enough to create the stroke gradient (2). Group this path with the one created in step 1. Create yet another linked offset path with no fill, and the same stroke width that the one in step 1, but with white colour (3) If you want to use a stroke style (dotted line etc.) set it for this path. Use the white stroked path as a mask for the group created earlier. Now you have a stroke with a gradient, and mostly it behaves like an ordinary stroke — edit the nodes of the base path, and the stroke changes with it. Even path operations work, as long as they produce only one path (if they produce several, the gradient stroke is only applied to the topmost one).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial4.png

[[Category: Art Tutorials]]

Create Art

2008-02-23T16:03:26Z

Sgt. Groovy: adding a tutorial

{| style="float:right; margin-left:1em;"
|__TOC__
|}

== Getting Started ==

=== General Information ===

Graphic artists usually meet on the [http://www.wesnoth.org/forum/viewforum.php?f=9 artwork development forum] or on the [http://www.wesnoth.org/forum/viewforum.php?f=18 restricted art development forum]. The former is a great place to post and discuss new and current Wesnoth art and graphics, and the latter to see what the art development team is working on.

Unit and terrain art is stored in the portable network graphics (PNG) format. Each frame of a unit animation, and each variation of a terrain is stored as a separate .png file in the "images" subdirectory of wesnoth, and generally these files will be 72 x 72 pixels (the size of Wesnoth's basic hexagonal tile) with an alpha channel (a part of the file that indicates how transparent each pixel is). When creating your own images, you can test them without overwriting any game data by putting them in your userdata directory (see [[EditingWesnoth]] for details on Wesnoth's directory structure). The game also supports JPEG images, though these are better suited for story art.

To edit these graphics, you'll need some program capable of creating PNGs - some of the programs in the following list are free, open-source software, and will do the job nicely: [[Art Programs]]

If you need some inspiration, we have a [[GraphicLibrary|Graphics Library]] which collects art posted on the forum. You can use this for ideas, and as a scrap heap for different parts of unit images (a technique described [[Give Your Hero A Personality|here]]).

=== Roadmap ===
A list of what's being done and what needs doing:
* '''[[Tiles Status]]''' - a roadmap/plan of sorts for future work on terrain tiles
* [http://www.wesnoth.org/forum/viewtopic.php?t=9162&sid=21cf168d2b077d06bb0017c059b4add6 A list] of current work that needs to be done with sprites.

== Art Tutorials ==

These are a work-in-progress, and describe both how to make art fit into wesnoth's style, as well as giving some considerable tips on drawing in general. Especially useful is the [[External Tutorials]] page which lists a large number of art tutorials available on the web.

=== General Art and Computer Graphics ===
* [[Using the Levels Adjustment]] - making scanned pencil drawings ''not'' look washed out.
* [[Extending dynamic range]] - The Grooviest (so far) tutorial about extending the dynamic range of images and how this technique can be used to make better scans of pencil drawings.
* [[Scanning with camera]] - How to transfer real-life art to computer using a digital camera instead of a scanner.
* [[How to Anti-Alias Sprite Art]] - A means of removing the jagged edges on pixel lines
* [[Art Supplies]] - What physical items you need to do larger cell-shaded art like that of Jetryl/Jormungadr/et al
* [[Inking With Pencils|Computer Inking a Sketch]] - Info from Jason Lutes on his portrait workflow
* [[Scaling Digital Images]] - how to properly resize an image on a computer
* [[How to Shade]] - at attempt at tackling a very complicated topic
* [[Cartography for Wesnoth|How to make Wesnoth-style Maps ]] - Kestenvarn's tricks of the trade
* [[Designing weapons and armour]] - Advice from zookeeper on designing realistic weapons for your characters
* Sgt. Groovy's vector workshop - tips and tricks for drawing with Inkscape
** [[Z-order tricks]] - few methods for faking overlapping shapes
** [[Variable-width strokes]] - how to make the strokes vary in width, like being drawn with a flat-tipped pen — no tablet needed!
** [[Shaped gradients with Gaussian blur]] - how to make gradients in other shapes than linear or radial
** [[Smooth shading in vector]] - the basic vector techniques for smooth shading, employing Gaussian blur and clipping/masking
** [[Making portrait art in vector]] - a complete tutorial for making Wesnoth unit portraits in vector graphics.

=== [[Terrain|Terrain Graphics]] ===

The following is information specific to drawing terrain for Wesnoth. Read Frame's "Tiles Tutorial" for a good overview of how terrain graphics work.

* [[Tiles Tutorial]] - Frame's tutorial describing the process of making terrain tiles in wesnoth, and how they interact with adjacent tiles.
* [[How To Make Seamless Tiles]] - The tutorial is aimed at Photoshop users, but the technique is similar with The GIMP.
* [[Seamless Tiles Using Inkscape]] - This tutorial teaches a method for making seamless hex tiles in vector craphics (to be rendered in raster).
* [[Turning Square Tiles into Hex]] - Nifty tricks for transforming square (or any rectangle) shaped seamless tiles into hexagon seamless tiles.
* [[CastleTutorial|Castle Tutorial]] - A description of how Wesnoth's castle tiles work (needs updating, but useful nonetheless)
* [[MultiHexTutorial|Multi-Hex Tiling Tutorial]] - A description of how multi-hex tiles work.
* [[Editing Castles]] - Instructions for how to make/edit castles (and other corner-based terrains) using yobbo's GIMP script.

These describe the system used to specify how terrains behave in game:
* [[TerrainCodesWML]] - A list of the letters used to represent terrain types.
* [[TerrainGraphicsWML]] - If you really need to get technical, start here.
* [http://www.anathas.org/ayin/wesnoth/doc/terrain_graphics_wml Ayin's Terrain Graphics document] - If you really, ''really'' need to get technical, this describes the terrain graphics WML system in depth.

=== Sprite Art ===
The following are different tutorials about sprite work compiled by various wesnoth sprite artists. These will give you the most specific-to-wesnoth information about making sprites, and are well worth a read.

* [[Creating Unit Art]] - a list of specifications you will need to match.
* [[Give Your Hero A Personality]] - tricks for editing existing images, including some clip art.
* [[Basic Animation Tutorial]] - or "How to Animate Sprites for Dummies," covering the basic theory, and all of the mistakes to avoid.
* [[Advanced Animation Tutorial]] - how to wire in the animation frames to units in a way that makes the animation look good in-game. Doesn't cover any drawing as such, but focuses on timing conventions and some standard techniques for getting as much as possible out of the frames that are already drawn.
* [[Team Color Shifting]] - how to use our new team color system.
* [[Creating Shadows Under Units]] - how we create the shadows for the units in-game
* [[Making Bow Animations]] - the current standard for how we want bow animations to work
* [[Rotate Pixel Art Without Blurring]]
* [[Creating a scratch built sprite]] - an attempt to show some ways creating a sprite from scratch
* [[How to create motion blurs]]- a simple explanation on how to create attack animation weapon blurs

=== External Tutorials ===
The following page contains dozens of links to tutorials covering all manner of artwork, including sprite art. These were not made by wesnoth artists, but should prove very useful for general instruction.

* [[External Tutorials]]

== See Also ==
* [[Create]]
* [[EditingWesnoth]]
* [[GraphicLibrary]] - Lots of usercontributed graphics (most should be GPL'ed)
* [[Project]]
* [http://wesnoth.dbzer0.com/blog/wpg2 External Graphic Library] - A project to better organize the art of Wesnoth.

[[Category:Art Tutorials]]

Smooth shading in vector

2008-02-23T16:02:38Z

Sgt. Groovy: Adding tutorial

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

In this tutorial we are going to learn few methods for making smooth shading in vector graphics. Shading is very important in bringing out the three-dimensional shape of objects, and combining Gaussian blur with clipping and masking brings much more power to accomplish it than mere gradients.

The important part with blurring to keep in mind is that the colour transition function is different than in gradients. Gradients have a linear transition (between any two stops), while in blurred objects it is S-shaped. What the difference means in practice is that even if a gradient goes all the way down to full transparency/background colour, it will still produce a distinct egde (top left in the image below). The blurred object, on the other hand, will have truly fuzzy edge (top right). It also makes a difference whether an object that is both blurred and transformed is first blurred or transformed. Bottom left is a square that has first been blurred and then scaled vertically down 50%. Bottom left is the same square, but it is first scaled and then blurred.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial7.png

The order is different when an object is both blurred and clipped/masked. In that case the blurring is always done first, and for that reason you can use clipping/masking to create objects with some edges sharp and some blurred. If you want the object to be blurred after clipping/masking, you have to put it in a group, and clip/mask the group (in Inkscape, you can't group single object, so you have to create a dummy object, group it with the real object, CRTL-select the dummy and remove it).

Let's start with a very simple example, a sphere. You probably already know how to do a shaded sphere with radial gradient, so here's another kind of method: First make a circle and fill it with the colour of the shadow side (1). Make a copy of it in the lighted side colour (lighter and more saturated version of the shadow colour), move it toward the light source (here up and to the left) and blur it (2). Make another copy of the original circle and use it as a clip path on the blurred circle. You'll see that the clipping cut off all the "leaking" blurriness, and the rest will make a nice smooth shading on the sphere. Let's make a shadow for the sphere too, while we're at it: First make a circle of the shadow colour and blur it (4) (it's most often a good idea to make the shadow as it looks from the above, and use transformations to bring it into right perspective). Real live shadows have the quirky property that their blurriness depends on the distance from the object casting them. The blur radius of gaussian blur filter is the same everywhere, but we can fake the effect with a mask. Make another circle, somewhat bigger than the shadow, white in fill colour and blur it, but less than you blurred the shadow (5). Use this as a mask on the shadow and you get nice variably blurred shape (6). With little scaling and rotating it looks good to go under the sphere (7).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial5.png

For a sphere you could have made a decent shading with radial gradient as well, but this approach works with other kind of shapes too. Sometimes you can also use the blurred path to make the shadow side, like in the image below.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial6.png

You can also use several blurred paths for shading, like below. The middle lightness (clipped) path was made by copying the original path, and the lightest one by making a linked offset (left). When both are blurred (with different blur radius) they result in a bit glossier surface than in the sphere example (middle). The technique works prettty well even for such odd shape, but it's not perfect. That's OK, there's nothing stopping you from editing the blurred paths, like has been done with the rightmost image (you will have to do "Path -> Object to Path" for the linked offset in order to node edit it). if you use a clone of the base path for clipping, you can still easily edit the base path after shading, because the clip path updates automatically.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial11.png

The above approach works best with smooth, organig shapes, but blurring can come handy when doing edgy shapes with flat shading. In the image below the edges are made smoother and more real life-like by a bit of blurring.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial8.png

Usually clipping/masking is uset to get rid of the "leaking" blurriness outside the path, but the "leaking" can also be used to one's advantage. In the image below there is a circle with a bigger one around it, a hole being cut in it's middle the size of the smaller circle. The bigger circle has been filled with a gradient (1). When the bigger circle is blurred, the blur will leak over the smaller one (2). When a copy of the small circle is used as a clip path on the bigger one, only the leak is left visible, creating a shading on the small circle (3). When a bigger blur radius is used, the shape changes from flat, button type one, to rounder one (4).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial9.png

Here's one more variation of these techniques, to produce a flat, paint splash type shape. First make the base path (1). Make a copy of it in darker, less saturated colour and move toward the shadow side (1). use a copy of the base bath to cut off ("Path -> Difference") parts of the shadow path and move back a bit (3). Blur the shadow path (4). Following the same procedure, create the highlight path (lighter more saturated, on the opposite side) (5). Clip both the blurred paths with a clone of the base path (6).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial10.png

==External links==

Here's some links to similar tutorials:
* More [http://howto.nicubunu.ro/inkscape_blur_orbs_3d_text/ blur shading] from Nicu (her other tutorials are also worth checking out)
* Awesome [http://www.inkscape.org/screenshots/gallery/inkscape-0.45-3D-rope.png "interactive" 3-d rope], demonstrating the power of clones
* Fairly complex [http://troy-sobotka.blogspot.com/2007/11/inkscape-tutorial-1-chrome-effect.html "chrome" text effect], where the text can be changed and the effect updates
* [http://chrisdesign.wordpress.com/2008/02/14/simple-metal-orb-using-gradients/ Realistic metal shading] made with multi-stop gradients

[[Category: Art Tutorials]]

Creating a scratch built sprite

2008-02-21T19:42:17Z

Sgt. Groovy:

== Intro: An attempt at spriting==
by [[user:Wayfarer alias Shadow|Wayfarer]]

Everyone seems to just "Frankenstein" sprites (ie. cobble them together from pieces of existing sprites). Now and then I'm asked how I create sprites from scratch.
To make it short (I hate long introductions) this is a kind of recipe, but it doesn't replace experience and practise.

'''Reguirements:'''
Graphic Program that can handle layers and transparency.
At least The GIMP, Photoshop. Pain.. err I mean MS Paint if you want to suffer and you will.

At least advanced knowledge about spriting. Sadly nothing for the beginner, sorry. You should have at least pushed some stuff around digitally.

==Anatomy==

OK after a while in the forums I realised that I missed one important point. Anatomy!

Shading, color choice, anatomy - it is all like music. Everything works together. Otherwise good art with wrong anatomy is like a nice voice with no rhythm.

Normal human anatomy is that legs are half the size of the whole body. In Wesnoth the legs are a bit shorter and the head slightly too big (That doesn't mean you can make it even worse - take another sprite and compare it to yours. If the anatomy differs too much you have made something wrong.).
That's the No. 1 reason I use capes - I try to disguise this anatomical flaw a bit. Nothing more nothing less.

That doesn't mean that sprites are completely detached from the real world. An awkward pose will fall into the eye of the viewer. So stick to the real world and imitate it. Use references!

In Wesnoth you'll probably mostly sprite warriors. You should note that most martial arts build on a firm stance. That means your sprite should show this.
Stretched Legs and not a guy that almost falls over or stands to attention. That doesn't mean that he doesn't look dynamic, far from it. Here are some good examples:

http://img525.imageshack.us/img525/3856/13tt0.jpg
http://img515.imageshack.us/img515/5781/11ck5.jpg

Note: Not done by me, linked from a Games Workshop artist. If there are any problems with them I'll remove them.

Everyone who has practised martial arts will know that this stance is stable - a good stable position with much room to move.
And the most important thing : it looks very dynamic.

I've used this trick quite successful on my latest sprites.
http://img248.imageshack.us/img248/7881/greatbowv2cg7.png http://img206.imageshack.us/img206/6696/slayerv14dl5.png

Let's conclude: simulate reality and take references. If you came up with a very difficult pose or you you can't put your finger on what's wrong with your sprite, look for real life pictures or refs from better artists.

I repeat it again: take a reference.
You can draw like Dürer, have Da Vinci's knowledge of colors your Picture still look like it has just escaped from Picasso's "Guernica". (Sadly all 3 were true for Picasso :( )

Other links provided by Jetryl:
Books about anatomy quite good and interesting for portrait art. The same rules apply to spriting too, but to a lesser extent.

[http://www.amazon.com/Human-Anatomy-Made-Amazingly-Easy/dp/0823024970/ref=sr_1_1/002-1709014-5120035?ie=UTF8&s=books&qid=1189548452&sr=8-1]Human-Anatomy-Made-Amazingly-Easy

[http://www.amazon.com/Drawing-Cutting-Edge-Anatomy-Reference/dp/0823023982/ref=pd_bxgy_b_img_b/002-1709014-5120035?ie=UTF8&qid=1189548452&sr=8-1]Drawing-Cutting-Edge-Anatomy

==Creating the base==

I started the same way as anyone else, creating sprites from cut and paste and smaller edits. But I was wonderfully free of any prejudices towards spriting like limited color schemes, hard edges or that the sprite must look like it is chiseled out of a rock. I was open to try something new and this has leaded to a bit quirky style.

Firstly, this means my art isn't always compatible with mainline but, well, you can't please everyone.
Secondly, I don't claim it is perfect - I just like it that way.

In the past I mostly did some Frankensteining to get the proportions right and the overall shape of the sprite until it looked like I wanted it. This is time consuming and actual very limiting. Some artists like [[user:Neoriceisgood|Neo]] seem to be able to draw a sprite just from scratch with a mouse.
I can't that's why I went another way.
Most of the time I make a small sketch of the sprite as small as possible to get an overview of how it will look. (Preferred 200 x 200 pixels)

http://img530.imageshack.us/img530/1092/wayfarerlinesdz6.png

Some experience with drawing is definitely helpful. The general anatomy isn't that important because Wesnoth sprites aren't that exact with anatomy anyway. Smaller flaws aren't too bad because they can be eliminated later with the coloring. The most important thing is to get the shape right and the look you want to achieve.
If you are happy with your first draft you can go to step 2.

Scaling down the pic. First crop the unused part of the sketch as possible then reduce the height or width (the one which is bigger) to 72 pixels.(cubic algorithm).
The result should be something like this.
http://img440.imageshack.us/img440/9718/linessmallax0.png

Next you should make another layer and lay some color under the lines (the flats)
http://img530.imageshack.us/img530/9875/linesflatqm2.png

As you can see it doesn't look very impressive but that is just the beginning. The most difficult part still awaits - the shading.

There's a different way to get to this stage that I used to prefer. In the past I filled some areas with colors and erased the edges until I got the shapes right (a little bit like carving). Occasionally I still use this technique for capes and other big flooding objects that should look random.
As I said that's my current way and it might not the best but it works for me at the moment (Yeah I'm still developing).
Experiment with it and find out what suits you the best.

Another example:
http://img240.imageshack.us/img240/8829/exebigrz7.png

That's the original size.
http://img530.imageshack.us/img530/1526/aragmagefemalese1.png
Cutting it down to the final size.
http://img240.imageshack.us/img240/3233/finalzc6.png

I used the same techniques for this one as described and it is actual inspired from a older drawing I did.
[http://fc04.deviantart.com/fs12/i/2006/306/b/5/The_Weight_by_wayfarerpg.png]Reference picture

I know I keep repeating: it is always helpful to have a kind of reference.

If you are good enough you can also sprite it in original size and draw it easily freehand (the other ways aren't that straightforward, they should just make the process a bit easier).
If you do this then you have 2 possibilities.
If I use a brush for the base shape I have the advantage that I can create the base without much erasing (at the beginning) and it comes very natural to me. Unfortunately, in the end I must hunt down dozen translucent pixels.

Pencil is a bit harder but I can avoid those artefacts that the brush causes. It boils down to personal taste.

I can do this too though it is just a matter of practice. Again there is no "one and only way". I can just show some possible ways. In the end it is just a question how much time you want to devote.

'''Example:'''

http://img229.imageshack.us/img229/6884/sorceressqi5.jpg
http://img409.imageshack.us/img409/1928/66863960kd6.png http://img409.imageshack.us/img409/8412/99517190fk9.png http://img248.imageshack.us/img248/2695/14990273go2.png

A quick hack that took me half an hour with the help of the reference. Still improvable but it should illustrate what I want to say.

==Color==

Okay, if I start with shading first a small course in color theory. This is more important for portraits but it can also be used for sprites.

The contrasts and color theory. I know what you think. http://www.adipositasgemeinschaft.de/forum/images/smilies1/blah2.gif

The fact is, colors can be ordered in a circle.

http://upload.wikimedia.org/wikipedia/commons/8/88/Farbkreis_Itten_1961.png

(Thanks to Wikipedia for a deeper analysis [http://en.wikipedia.org/wiki/Color_theory Color theory]

although the German site is better - those who can read it should do so
[http://de.wikipedia.org/wiki/Simultankontrast#Simultankontrast_-_Sukzessivkontrast Kontraste])

So there are 7 different kind of contrasts I'll try to at least briefly cover all of them.

The first is the '''complementary contrast''' . If you look at the circle, orange lies opposite to blue. That means both colors painted next to each other will create a strong contrast. It's an eye-catcher but be careful as it can make the sprite quite noisy.

The second '''dark and light'''. A good example is missing, I'll add one later. Many dark areas and only some brighter areas.

'''Cold Warm contrast'''
People find orange and red as "warm" while they associate blue with "cold" this can be used too but should be used sparingly.

'''Color contrast''' or how is it called well Bogus for us we use colors.
Example
http://upload.wikimedia.org/wikipedia/commons/1/18/Marc-blue-black_fox.jpg

'''Simultaneous contrast - successive contrast'''
Well now it gets interesting. Neo uses this from time to time to add more depth to his shading.
http://upload.wikimedia.org/wikipedia/commons/6/6e/Simultan2_Kopie.jpg

He lays pixels in different colors next to each other so they boost each other.
For example on normally white highlights bright blue or green. Surrounded by other colors.

The '''Quantity and Quality contrasts'''.
Drum roll, please.

No, seriously - some big words, but in fact the first is easy. Quantity means, for example, that a big picture is mostly green and with some small red areas. They will fall in the eye of the watcher.
The next one is the most important. Lucent pastel colors (mostly mixed with white) next to blunt and dark colors (grey and black mixed in).

A good example is this painting of my favorite artist.
http://upload.wikimedia.org/wikipedia/commons/thumb/d/dc/Caspar_David_Friedrich_035.jpg/800px-Caspar_David_Friedrich_035.jpg

To conclude I use mostly Quantity and Quality contrasts and dark and light contrasts. I'll show soon how I put them on a sprite.

I've now probably chased away all my readers with my http://www.adipositasgemeinschaft.de/forum/images/smilies1/blah2.gif.

But these are the tools that an artist can use to make his art ''good'' art.
It's not like I say "well now I use Cold Warm contrast". I experiment until I think it looks right. It is just helpful to know the tools you have at hand.

Further external links kindly provided by jacobolus. For a bit deeper overview.

[http://en.wikipedia.org/wiki/Munsell]color theory

[http://handprint.com/HP/WCL/color11.html]color theory again

[http://handprint.com/HP/WCL/wcolor.html]and again

Though this gets more interesting for portrait art.

==Detailing==

In 25 inconvenient steps to a scratch build sprite. Only $5, a bargain price soon in your library.

Just kidding.

http://img410.imageshack.us/img410/4712/testtl2.png

1.
I did pretty much know what the sprite should look like. For example I can take a ref picture and I can produce a basic sprite out of it. It gets harder if I don't actual know what I want. In that cast I can waste hours on it. (That means a good ref picture is priceless, even a quick sketch you made for yourself helps).
It helps immensely if you already know what you want.
I use different approaches to create my sprites. Sometimes I fill large areas with color and erase the parts that look wrong. Sometimes I make a small sketch (as described in part 2). Mostly I just mix it. Like here. (The different sprite at the end is just there for size comparison purposes).

I should note that I normally Frankenstein some parts to speed up the process. I mustn't draw a new face just because they look all very different from each other. http://wesnoth.org/forum/images/smiles/icon_wink.gif
In this case it is completely from scratch.

(Other example partly with Frankensteining to show how short you can cut it)
http://img162.imageshack.us/img162/6538/progressgo5.png

That means I'm not against Frankensteining - it just limits the possibilities.
Nothing more nothing less.

As you can see I struggled from 1 till 2 to get the shape right. Normally you mustn't start to make the basic shading until the shape is right. I do it anyway because I can't work until I get a feeling how the finished sprite will look. (Everyone has his own small quirks) http://wesnoth.org/forum/images/smiles/icon_rolleyes.gif

Anyway, once you have the shape right you should apply basic shading. This consists of at least 3 different shades of the colour. With these colours you start molding the highlights, the shadows and the textures on the folds.
As you can see here in big.

http://img441.imageshack.us/img441/3483/exeqn7.png

Dark areas with darker colours. Exposed positions, uprisings with bright colours.

As the cretin I am I can't count to 5 - I've forgotten to add 3.
Anyway, I added the hair and more details until I reached stage 5.

This is nothing too fancy. The only thing I can say about it: keep everything that covers the body in different layers. In my case the book and the staff (I have switched them various times to see where they fit best).
This is for 2 reasons. Firstly, it saves you time and makes it easier to animate the whole thing. Secondly, if you don't like it you don't have to erase half of the sprite and redraw it. BTW I learned this the hard way too.

5. Actually, doesn't look bad does it? (At point four I added some green to the white highlights to let them stand out a bit, like Neorice does.)
If you are good you can work further with a pencil tool and further shades.
If you want it different, well:
First I use these tools: sponge, dodge and burn.

http://img207.imageshack.us/img207/7158/toolsab9.png

(Dodge and burn are available in both The GIMP and Photoshop - sponge is only available in Photoshop). You should test out the different settings to see which one fits you the most).

These tools are not essential - they just make it easier. You can emulate the effect with carefully chosen brush strokes. If you don't have brushes, well, sucks to be you. ;)
Seriously, you can emulate it with careful chosen pencil dots. This goes much too far for me, personally.

Basically tools that adjust the brightness and contrast.
In the first round I add darker areas. I deepen the shadows or add further ones. Than I make the highlights brighter. I use these tools on large areas to get a nice smooth transition between the different areas.

Before and after

http://img512.imageshack.us/img512/163/93794620nu1.png http://img514.imageshack.us/img514/7233/35075772dk7.png

Than I work on the finer shadings with a pencil with low opacity and fading out.
I add reflections and make the shadows smoother in the process. Mostly with white and black, sometimes with other colours. Because the reflection of light is not always white and not every shadow is just black. Again real life gives nice sources of inspiration. (Golden reflections on blue textures. Red on polished metal, etc... use your imagination)

'''A big note though!!!'''
Shading <> ambience light. Many seem to misunderstand it. Shading means that you model the surface. You add folds to cloths, waves to hair and surface to faces. You make the sprite dimensional. Ambient light is a light source that adds different light values onto a surface to make it more interesting. Ambient light without proper shading doesn't work but it goes the other way around too. A sprite without prober ambient light looks dull.

Before and after
http://img530.imageshack.us/img530/5194/49958573tb5.png

I probably went overboard with the detailing in the process but magnified you quickly lose the overview. That's why you should keep a window open displaying the sprite at original size. Test the sprites very often against in-game background. Else you might get evil surprises.

To conclude:
-Shaded surface.
-Worked out details
-defined light source,ambient light
-smooth transitions avoiding the "comic" effect(areas separated by lines, you know those cheap comics in newspapers)

Another thing is TC (Team colouring). I don't really like it because I lose control of how the colours of the unit will work together with TC. That, and the fact that I can't just paint it the way I do the rest of the sprite. This means you keep it best in a separate layer too (also, if your brush slips it won't work anymore happened to me too, actually far too often). I changed it various times until I found the right position for it and pink fits to nothing. http://wesnoth.org/forum/images/smiles/icon_lol.gif

This is basically one way, but there are various variations of it. (For example brushing larger areas of the cape, which is a bit more trickier, because you add the folds on the fly with a brush and you'll need a calm hand to avoid the straight lines). Basically, everyone develops their own way which is the easiest for them. And no, there is not the one and only way - just various paths. :)

'''Further links:'''

[http://www.warpedspace.org/lightingT/part1.htm]light and reflection

'''Internal links:'''

[http://www.wesnoth.org/wiki/How_to_Shade] How to Shade

[http://www.wesnoth.org/wiki/How_to_Anti-Alias_Sprite_Art] How to Anti-Alias Sprite Art

[http://www.wesnoth.org/wiki/Team_Color_Shifting] Team Color Shifting

==Conclusion==

Don't do. (I tell you)

* Use rough edges. Straight lines are evil. 90°, 45° degrees too.
* Symmetry is boring, a symmetric face stands for beauty but still boring.
* Avoid the "chiseled out of a rock" look (moved cape stretched stance,.. etc help avoid that).
(A doctrine of the first greek sculptor was that the statue can be rolled down a hill without any damage. Don't make the same mistake)
* You like dark colors, fine. But make sure you have enough highlights in it, so the unit doesn't looks washed out or boring.
* Don't expect details which are 2*2 to be seen. Only if the viewers second name is "Hawk Eye".
* Don't start with a highly complex project and give up frustrated in the middle of it
* Don't start emulating some anime stuff you have seen. Anime exaggerates too much and breaks every physical law.
* Don't separate the different parts of a sprite with a 1 pixel wide line ie. the "comic" effect. Don't do it. If you must, let it look like a shadow a 2 or 3 pixel wide line of darker shades. Play around but avoid this 1 pixel line because the line will have a high contrast to the other color areas and if it separates two different color areas from each other it looks ugly.
* Don't overreact to critiques. Everyone makes mistake and the one on the other side of the monitor isn't be always right.

Do
* Monochromatic can work if the contrasts are high enough.
* Show stuff moving - use wind as another artistic tool.
* Use optical illusions. If you can draw a face with 4*4 pixels fine but no one will see it. Hint it.
* Choose one light source and stick to it.
* Small steps if you are a beginner but steps and not backwards is not the right direction.
* Do good taste a favour and avoid color overload. aka Rainbow warrior (You can use many colors but stick to some major colors and add the other colors in a limited amount).
* Exaggerate slightly if it helps to communicate the message.
* Realism is good. Try to stick to it but not so much that it would hinder you.
* If you paint a unit take a break from the brush tool and use the pencil and add some straight lines, especially on the edges of capes to break up the shading. It can add some very nice contrast. Use it with care. (On texture it looks like embroidery)
* Use smooth transitions between the different parts of the sprite.
* If you need the space take it. You won't achieve anything with 2*2 pixels
* Accept critique. Actual learn to critique yourself.
* Experiment. Try something new.
* Save often. Look often at the real size. Look at it ingame.
* Put it away and make a new start after some sleep.
* Keep copies of various development stages. Once you made a mistake you can still give the whole thing a new direction.

So Questions anyone? Suggestions? Random praises? Flames? (Warning: I flame back).
So feedback can be posted in the forum in the appreciate topic. I'll try to work the feedback in.
[http://www.wesnoth.org/forum/viewtopic.php?t=16760&postdays=0&postorder=asc&start=0]

To close it with one of my favorite philosophers:

''I know that I know nothing. You pay for a round? A double one for me!''

'''Note I'll try to replace the example pics with something under GPL to avoid any trouble. At the moment I haven't much better stuff.'''

[[Category: Art Tutorials]]

Advanced Animation Tutorial

2008-02-21T19:40:41Z

Sgt. Groovy:

== How to write animations ==

This page deals with how to write the animation WML when you got the frames drawn and ready as well as general guidelines for animation lengths and timing. Proper timing is in many cases almost as important in getting an animation to look nice, smooth and consistent with the rest of the game as the actual frames are.

=== Timing ===

As a general rule, a melee swing should last at least 300ms and at most 400ms: about 200ms before and after the moment when the swing hits, to keep the speed at which the unit slides back and forth somewhat constant. Naturally there can be exceptions to the "normal" animation lengths, like woses, which have much slower attack animations than most. With ranged attacks, the length of the animation isn't quite as important, since the units aren't taking turns in sliding back and forth (as is the case with melee combat), which makes different animation speeds much more visible. With ranged attacks, however, you have an additional animation to consider: the missile itself. Most normal missiles such as arrows, knives and spears should fly for 150ms. Magical missiles, big rocks or other unusual projectiles can fly longer, but 150ms is the unofficial standard for normal missiles.

=== Filler frames ===

The most common use of this is when you only got one frame to represent an attack. Usually, a single attack frame sliding back and forth (if it's a melee attack) looks pretty bad. Ideally, you'd want to have an animation that has the unit stepping forwards and backwards while swinging his weapon, but now we'll assume that for some reason you have to do with just one drawn frame.

Instead of using the common method, which looks like this:

[animation]
[frame]
begin=-100
end=100
image=unit-attack.png
[/frame]
[/animation]

...you can with little trouble make that attack smoother, by having the attack frame kick in only at about halfway into the attack and also having it last only slightly past the point where it actually hits the enemy. How do you do this, if you only got one frame to work with? Well, the easiest solution - and one that works fine with almost any kind of unit - is to use the base frame:

[animation]
[frame]
begin=-200
end=-150
image=unit.png
[/frame]
[frame]
begin=-150
end=100
image=unit-attack.png
[/frame]
[frame]
begin=100
end=200
image=unit.png
[/frame]
[/animation]

This makes most melee attack frames look much more natural, since this way it seems more like the attacker is hitting the enemy he's approaching, instead of hitting air where he stands and then sliding towards the enemy and back. The same technique can and should be used whenever you have less than 3 drawn attack frames, whereas animations that have more than that would probably benefit very little from this.

=== Simple way of getting new drawn frames ===

Sometimes, the unit has frames made for other animations that can be used as part of the attack animation. Most usually this is a defense, movement or a leadership frame. Quite a few defense frames, for example, are quite suitable as a wind-up to the attack.

...

=== Sounds ===

For the vast majority of attacks it is important that it has different sounds for when the attack hits and when it misses. Especially important are "normal" weapons, such as swords, bows, spears and so on. Magical attacks don't necessarily need a different sound when they miss, but if one is available, it should naturally be put to use. Absolutely never leave an attack without any sound altogether! If there isn't a fitting sound available for that weapon or attack, use the second best fitting sound. No attack, unless it's by a silent psychic assassin perhaps, should ever be without an accompanying sound.

One common way to create hit and miss sound variations is by creating two separate ''[animation]''s for the attack and filtering them by the ''hits='' condition, like this:

[animation]
hits=no
[frame]
begin=-200
end=-150
image=unit.png
[/frame]
[frame]
begin=-150
end=100
image=unit-attack.png
sound={SOUND_LIST:MISS}
[/frame]
[frame]
begin=100
end=200
image=unit.png
[/frame]
[/animation]
[animation]
hits=yes
[frame]
begin=-200
end=-150
image=unit.png
[/frame]
[frame]
begin=-150
end=100
image=unit-attack.png
sound=axe.ogg
[/frame]
[frame]
begin=100
end=200
image=unit.png
[/frame]
[/animation]

The other way is to use ''[if]'' and ''[else]'' tags in the animation. In this case, they could be used like this:

[animation]
[frame]
begin=-200
end=-150
image=unit.png
[/frame]
[if]
hits=no
[frame]
begin=-150
end=100
image=unit-attack.png
sound={SOUND_LIST:MISS}
[/frame]
[/if]
[else]
hits=yes
[frame]
begin=-150
end=100
image=unit-attack.png
sound=axe.ogg
[/frame]
[/else]
[frame]
begin=100
end=200
image=unit.png
[/frame]
[/animation]

In case you're wondering, the ''{SOUND_LIST:MISS}'' is a simple macro that just expands to ''miss-1.ogg,miss-2.ogg,miss-3.ogg''. A comma-separated list of sounds can be given to all ''sound='' keys, making the game pick one of them at random whenever the sound is triggered, and the macros (you can find them in ''data/core/macros/sound-utils.cfg'') exist simply to make writing them more convenient.

=== Slows, poison and other special situations ===

Attacks with the slows or poison specials should emit appropriate sounds when they successfully slow or poison an enemy. We have two ready-made macros, FRAME_ON_SLOW and FRAME_ON_POISON available for this which should make the task trivial. Below is an example of how FRAME_OF_POISON is used on mainline units (also note the use of the above-described HIT_MISS_SOUNDS):

[animation]
[missile_frame]
begin=-150
end=0
image="projectiles/dagger-n.png"
image_diagonal="projectiles/dagger-ne.png"
[/missile_frame]
{HIT_MISS_SOUNDS throwing-knife.ogg throwing-knife-miss.ogg (
[frame]
begin=-200
end=-100
image="units/orcs/assassin-ranged1.png"
[/frame]
)}
{FRAME_ON_POISON (
[frame]
begin=-100
end=0
image="units/orcs/assassin-ranged2.png"
sound=poison.ogg
[/frame]
)}
[/animation]

The FRAME_ON_POISON macro makes whatever sound you specify inside the enclosed ''[frame]'' (so you could give different poisoning sounds to different units, for example to make magical poisoning sound different from traditional poison, like that on poisoned knives) to be played when the attack hits and the target is not already poisoned, with no sound played if the attack misses or the target is already poisoned. The FRAME_ON_SLOW macro is used in an identical manner, except of course with a different sound (slowed.ogg) as the standard one to use.

[[Category: Art Tutorials]]

Creating Unit Art

2008-02-21T19:40:08Z

Sgt. Groovy:

Here's the information needed to start doing units art for Wesnoth. Please update the wiki with links to examples and other tips and tricks realted to unit art. This page started from [http://www.wesnoth.org/forum/viewtopic.php?p=104809#104809 this forum thread].

=== Basic unit image specifications ===
*Final format: transparent PNG, 72 x 72 pixels
** 8 bits ''per'' RGB channel. may be listed as "PNG-24" or "8-bit RGBA".
**"Indexed color" or "color map" is different. These will work, but are not recommended.
*the units should be contained in the hex (the white in the attached template)
*Normal Units (humans and elves) are about 40px tall
*Small Units (dwarves and goblins) are a bit over 30px tall
*Tall Units (drakes and great trolls) are a bit under 50px tall
* note: standard unit heights are in flux
*Light comes from a specific direction (for shadings and highlights)
*Unit must look in the lower right direction
*Units are centered horizontally
*Unit's feet are positioned around 55 pixel from the top, lower for taller units
*Shadows at 60 opacity, black color (with blurred and soft borders, see [[Creating_Shadows_Under_Units]]).
*Outlines should be used, made with a darker, but similar colour to what they're surrounding, this makes the unit feel bigger then the amount of pixels you'd use for the same with a black outline. (Black outlines not welcome anymore)
*Use consistent names for your files, beginning with the unit name, so that they are kept together when browsing the files.

=== Unit animation specifications ===
*You can have as many frames for animations as you like. It depends on what you want to get accross
*Images and animations must be done for:
*normal image (1 frame)
*attack for each type (at least one frame - 2 is good - 4 better)
*directional attacks (spear and ranged)
*defense
*death frame
*magical attacks should have 'special effects'

You can also have multiple defense animations (like the thief.)
Defense can be split into ranged and melee animations.

It is important that the unit moves its feet when doing a melee attack. During the attack the unit will move from its own hex to the hex it is attacking. If you don't have feet movement in your animation, the unit will be "sliding" toward it's target, which looks bad.

=== Additional Notes and examples ===

*Test early with different backgrounds. Go to your game terrain files, pick, say, 3 terrains, preferably with different colours, and have a look what they look like against the background, because that way if you have used alpha channels, it may look better in game than as the average standard sprite.

*Test your unit early in the game. Replacing an existing unit graphic with yours is the fastest way (be sure to make a copy of the originals so you don't lose them.)

*If you are animating a horse, you should base it on this [http://www.wesnoth.org/forum/viewtopic.php?t=7592&sid=eb882602a04b48cf8fcce9ca77fbeae0 example of an horse animation]

*The thief is a very good example of what can be done. It has basically all possible animations. It doesn't use the multidirectional ranged images, though, since it's not a ranged unit.

== See Also ==
* [[BuildingUnits]]
* [[Create]]
* [[Create Art and Music]]
* [[Art Tutorials]]

[[Category: Art Tutorials]]

Editing Castles

2008-02-21T19:37:37Z

Sgt. Groovy:

This tutorial is based on a gimp-python script, and as such is aimed squarely at users of The GIMP.

It will show you how to edit the existing castle tiles while making sure they continue to line up properly in game. The same technique can be used for any tiles which work on the same principle (such as Eleazar's new canyons, or Scott's indoor walls).

'''You will need to download both [http://www.wesnoth.org/forum/download.php?id=5197 this overlay template], and [http://www.wesnoth.org/forum/download.php?id=5196 this gimp-python script]. ''Extract the python script to your GIMP "plug-ins" directory'', and you're ready to go.'''

== Editing the Existing Castles ==

=== Step 1: Open ===

Open the overlay template in The GIMP.

http://exong.net/wesnoth-attach/files/shot1_568.jpg

Find the castle tiles you want to modify and drag them onto your image. This should add each castle image as a separate layer. In this case I'm modifying the "castle-convex" parts of pekka's castles.
Press 'M' and move the layers around so that the images line up around the central white hex. The layers should snap to the guides in the overlay template, so this should be easy to get exactly right.

http://exong.net/wesnoth-attach/files/shot2_531.jpg

You can hide the guides (as I've done here) by pressing shift-control-T, or deselecting "show guides" in the "View" menu.

=== Step 2: Edit ===

Edit the castle to your heart's content. Here I'm just going to put some annoying little flags on top. Make sure you edit one corner at a time, and that you keep each corner on a separate layer. It's useful to turn "Snap to Guides" off (in the "View" menu) if you're doing any freehand drawing.

http://exong.net/wesnoth-attach/files/shot3_178.jpg

=== Step 3: Save ===

Now to save the edited castle. First, make everything except for one corner invisible.

http://exong.net/wesnoth-attach/files/shot4_494.jpg

Then run the "WesCutter" script. If you've extracted the plugin to the right place, it should be found under "Wesnoth" in the Gimp "Python-Fu" menu. You'll be doing this again, so click the dotted line to detach the Wesnoth menu and save yourself a few clicks.

The script lets you choose a directory to save the file in (I'm saving to my desktop), the base filename (in this case, "castle"), whether it's a concave or convex image (added to the filename) and which corner it is (in this case, NW).

http://exong.net/wesnoth-attach/files/shot5_193.jpg

After clicking on "OK", I have a shiny new image called "castle-convex-nw.png" on my desktop, just the right size to work in Wesnoth. After doing this with all 6 corners, you can put them in the "images/terrain" subdirectory of your userdata directory (see [[EditingWesnoth]]). If you've put them in the right place, your new castles will automatically replace the old ones next time you run wesnoth http://www.wesnoth.org/forum/images/smiles/icon_smile.gif.

http://exong.net/wesnoth-attach/files/shot6_273.jpg

hooray!

== Making other Corner-Based Terrains ==

The same technique can also be used to make other corner-based terrains (for examples of other corner-based terrains, see [http://www.wesnoth.org/forum/viewtopic.php?t=7936 Eleazar's Canyons] and [http://www.wesnoth.org/forum/viewtopic.php?t=8274 Scott's Indoor Walls]).

== See Also ==

* [[Create]]
* [[Create Art]]
* [[CastleTutorial]]
* [[EditingWesnoth]]

[[Category: Art Tutorials]]

MultiHexTutorial

2008-02-21T19:37:18Z

Sgt. Groovy:

(quickly written, I'll explain it more in detail when I have more time -- Ayin)

== First example: mountains ==

Suppose we have the following map to fill with mountain tiles:

http://www.anathas.org/ayin/wesnoth/tutorial/tutorial1.png

And we have the following images to fill it:

* tile1: http://www.anathas.org/ayin/wesnoth/tutorial/tile1.png
* tile2: http://www.anathas.org/ayin/wesnoth/tutorial/tile2.png
* tile5: http://www.anathas.org/ayin/wesnoth/tutorial/tile5.png

We will use the following three rules to fill the mountains. Beware: '''Spaces into maps do matter!'''

[terrain_graphics]
map="
. 1
.
. 1
.
. 1
1
1"
[tile]
pos=1
type=m
[/tile]

probability=30
no_flag="base"
set_flag="base"

[image]
layer=-1000
name=tile5
[/image]
[/terrain_graphics]

[terrain_graphics]
map="
.
1
1"
[tile]
pos=1
type=m
[/tile]

probability=30
no_flag="base"
set_flag="base"
[image]
layer=-1000
name=tile2
[/image]
[/terrain_graphics]

[terrain_graphics]
[tile]
x=0
y=0
type=m
[/tile]
probability=100
no_flag="base"
set_flag="base"
[image]
layer=-1000
name=tile1
[/image]
[/terrain_graphics]

The first rule, regarding the biggest mountains, will apply first. It will include, with 30% probability, the big
5-tile mountain range. This will make, for example, something like this.

http://www.anathas.org/ayin/wesnoth/tutorial/tutorial2.png

The second rule will then apply. Note that the previous rule had set the "built-m" flag, thus preventing the mountains
from the second rule to overlap the place where the first mountains already have been placed. The result will be
something like that:

http://www.anathas.org/ayin/wesnoth/tutorial/tutorial3.png

The, the last rule, with 100% probability (the probability key could have been omitted) will apply. It will place a
mountain tile on the places where no other mountain tile was placed. The final mountain range will look like that:

http://www.anathas.org/ayin/wesnoth/tutorial/tutorial4.png

== Village with a tower ==

To add a building with a tower that spans over 2 hexes, you may use the following rule (T being the character
representing the building hex)

[terrain_graphics]
map="
**

T"
[image]
name=tower # a 36x72 image
position=vertical
base=36,108
[/image]

[/terrain_graphics]

== Reference ==

Map:

'.' means "this rule does not apply to this hex" (i.e: no image will be set to this hex, this hex will not be checked
for flags, etc).

'*' means "this rule applies to this hex, but this hex can be any terrain type".

Flags:

no_flag="built-m" means that the rule only matches if the flag "built-m" is not set.

set_flag="built-m" means that the rule, upon matching, sets the flag "built-m".

== See Also ==

* [[BuildingRuleWML]]

[[Category: Art Tutorials]]

CastleTutorial

2008-02-21T19:36:58Z

Sgt. Groovy:

== Building tileable castles - A tutorial ==

== Introduction To Castles ==

===Overview===

Wesnoth terrain graphics system allows the WML designer to create various
terrain graphics "models": rules that allow to determine how the different
terrain tiles will be displayed by the game. Those allow the terrain designer
to specify renderings that go beyond the simple "one terrain type = one image"
representation.

One of those terrain models is the Castle (it will be called like that,
"Castle", all along this document, although this model can be used for many
other terrain types, like, for example, caves.) Castles are created from two
type of elements: a castle floor, and a castle wall. Castle floors are normal,
basic tiles that are placed on each corresponding tile. As those really have
nothing special, they will not be detailed in this document. What we will
detail is how to build castle _walls_.

===The Simple Castle Model===

The Simple Castle Model is used to build castle walls, without keeps. Castle
walls are special transitions, that are drawn between Castle tiles, and
Non-Castle tiles. For example, imagine some castle tiles are disposed as shown
in figure (1); castle walls will then be drawn as shown in figure (2)

''(figure 1)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure1.png

''(figure 2)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure2.png

Castle Walls are composed of 12 different images (as show in figure 3), that
tile together. Those are positioned at the corners of the castle border, as
shown in figure 4.
"convex" images are positioned at convex borders, "concave" images are
positioned at concave borders (as it hopefully might be expected.) When we talk
about "concave", or "convex" borders, we always take the castle's interior
point of view.

''(figure 3)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure3.png

''(figure 4)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure4.png

===The Keep Model===

The Keep Model is created to build the special transitions that are used to
build a wall around the Keep of a Castle. Keeps are assumed to be always
composed of one tile (isolated from any other keep tile.)
Keeps are a little more complex than castles walls: they are built from 24
different images (see figure 5,) thus setting to 36 the number of images
required for a complete castle.

''(figure 5)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure5.png

There are 4 kinds of keep images: those representing keeps corners that are
outside of a castle, those representing keep corners that are inside of a
castle, and thos representing keep corners which are in the border; the last
one being divided in 2 categories, depending on the castle-keep tiles going
clockwise, or counter-clockwise (see figures 5 and 6.) Interior keep tiles are
supposed to tile with other interior keep tiles, and with corresponding
transition tiles; same for exterior keep tiles. Figure 6 illustrates the
different types of transitions for keep tiles.

''(figure 6)''

== The Wall ==

===Tile drawing problematics===

To draw tileable castles, one should start with drawing the wall. Some tileable
castles may only consist on a wall, without a keep (Think here that what we
call "Castles" in this document may apply to a lot of other terrains, built
with the same rules. Also think that some castle may have keeps that are normal
tiles.)
The 12 tiles that compose a wall are made from 2 groups of tiles, as stated
above: convex wall tiles, and concave wall tiles. Tiles for each one of those 2
groups are drawn together, then automatically cut out using a specialized
software. That way, only 2 graphics are necessary to generate those 2 tiles.

The main problem, as that should be inferred from reading the tile model
description, is making convex graphics tiling with concave graphics: as concave
graphics are drawn together, there is no problem making them tile together;
same with convex graphics. See figure 7 for an illustration of this problem: ||A and
B are drawn together, so they tile perfectly. However, we must make sure A will
correctly tile with C!||

''(figure 7)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure7.png

Geometry, and a good methodology, may be used to make those tile correctly.
This is possible, but, trust me, this is tedious. And hard to maintain. And
even harder to explain. And, the more complex the graphics are, the harder it
is. That is why ia helper tool was developed, to help in the process.
The helper tool we are talking about is called the Castle Exploder. We will
describe how to used it to build a complete castle.

===Preparing the work environment===

Before using the Exploder, you must prepare a work environment. The Exploder,
must be provided with files named according to a correct pattern in order to be
able to work.
Exploder files are named according to the type of image they represent. The
"type" of the image corresponds to the position of the different tiles that
compose the image. For example, the exploder recognizes any image that is
called foobar-convex.png as containing the 6 convex castle walls, disposed
around an hex.

There currently are 7 different image types; all of which are used to build a
complete castle with keep. Those are named:

* convex
* concave
* exploded-convex
* exploded-concave
* keep (not yet available)
* keep-walls (not yet available)
* keep-flower (not yet available)

''(figure 8)''

Forget about the last 3 ones for now; the 4 first ones are used to build
castles.

Download the castle project here:

http://www.anathas.org/ayin/wesnoth/castle-tutorial/files/castle_project.zip

To prepare your work environment, unpack the castle_project.zip file in a
directory of your choice.

===Drawing The Convex Wall===

Open, using your favourite bitmap editor the file "castle-convex.png" (if your
favourite bitmap editor cannot open png files, I can't help you.) You will see
the following:

''(figure 10)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure10.png

Do not be afraid with all those coloured polygons! Those just are guides, that
will help you to position your image. If your bitmap editor handles layers, I
suggest you make the actual drawing on another layer, keeping this one to
provide guidelines. If it does not, you should just delete all this coloured
stuff.
The different hexes, corresponding to the position of the actual terrain hexes
in the game, are represented with a value change. The 6 different corners, that
will be cut into tileable pictures, are represented with a hue change.

This image is supposed to represent a convex castle tile: that is, a single
castle hex, followed by non-castle hexes all around. Draw the corresponding
graphics on it. You do not have to bother about them tiling together: they
will, as they were drawn together.

Some guidelines on drawing the wall: when drawing a wall, the TOP of the wall
(not the bottom) should be around the hex border (TODO: confirm this: this is
good for the positioning of units, but not so for villages near castles)
Ideally, a wall should not be higher than 1/4 of a tile (that is: 18 pixels).
Try to make the castle appear to be viewed under the same angle as other
castles, and villages.

Let's draw the wall (figure 11).

''(figure 11)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure11.png

===Exploding The Wall=== http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/castle-exploder.png

Now, we must draw the concave version of the wall. To make sure it will
correctly tile with the convex version, we will not directly draw it in the
castle_concave.png image: we will draw it in an exploded version of the
castle_convex.png image. That is where the Exploder Tool comes into play.

Just open a terminal window, and type the following command:

exploder castle-convex.png castle-exploded-convex.png

The exploder will open the first file, extract from it the tiles it
recognizes, then put those on the second file, if there is a place to put
them (For example, exploding castle_concave.png to castle_convex.png is a
no-op, as those two pictures do not have any tile in common.)

''After the program finishing merging the tiles'', the castle-exploded-convex.png
file will then look like figure 12

''(figure 12)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure12.png

===Drawing The Concave Wall===

Draw the castle graphics on the 6 concave tiles, so that they tile correctly
with the convex tiles. It is useless to modify the concave tiles; if you have
inadvertently done so, you may explode back the castle_convex.png file on the
file you are working on. Do not worry: your work will not be lost; as the
exploder tool does only overwrite the portions of the image it actually
replaces.

''After you have drawn the missing castle corners'', the castle-exploded-convex.png
file should look like figure 13

''(figure 13)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure13.png

===Imploding Back the Concave Wall===

The concave tiles you drew should, magically, tile together. However, you may
want to ensure they actually do, and, if they don't, to correct them so they
do. Just implode back the castle_exploded-convex.png file into the
castle_concave.png file, using the command

exploder castle-exploded-convex.png castle-concave.png

It should look like as in figure 14.

''(figure 14)'' http://www.anathas.org/ayin/wesnoth/castle-tutorial/images/figure14.png

===Further manipulations===

===Cutting The Tiles For Use In The Game===

Once you have up-to-date castle_convex.png and castle_concave.png files, you
must cut them, so they are usable by Wesnoth. To do this, create a directory
where you want the cut tiles to be output (or you may just use the "output"
directory which already was created), and use the commands:

cutter castle-concave.png output
cutter castle-convex.png output

== Keeps ==

As it was said above, keeps are a little bit more complicated than castles.
Keep always being composed of one isolated tile, keep tiles always are convex,
however, they may have several different relationships with nerby castle tiles:
they may be inside of a castle, outside from a castle, or in a border.

===The keep views===

===Drawing interior and exterior keeps===

Keeps inside of a castle, and keeps outside of a castle, may be strictly
identical. However, for your graphics to be more "realistic," you may decide they
are different. For example, keep tiles located outside of a castle may have no
doors, less windows, they even may have a stronger wall, etc. Interior keep and
Exterior keep tiles do not need to tile together, as there always will be an
intermediate Border tile to make transition. They may be made separately;
however, keep in mind that you will have to make each transition between those.
If they are too different, transitions may be clunky.

===Making keep transitions===

===Flower Power===

[[Category: Art Tutorials]]

Turning Square Tiles into Hex

2008-02-21T19:36:05Z

Sgt. Groovy:

In Wesnoth the Terrain graphics are made with seamlessly tiling hexagonal pixmaps. These are more difficult to make than square tiles, for which ready made scripts exist in most image manipulation programs, and which can also be found in multitude of libraries, free to use.

If you cut a square tile into a hexagon, it's only going to be seamless in north and south edges. To make it seamless in all edges takes a bit of trickery, but it's relatively simple. There are two methods to do it, the one that is better depending on the circumstances.

==Method 1: Keeping the Orientation==

Let's start with a dry mud texture from the standard GIMP library:

http://koti.welho.com/thonkasa/Roinaa/hextile1.png

First it must be scaled into 54x36 pixel size (this applies to the standard 72x72 pixel Wesnoth hexes, for other hexes the figures are width*0.75 and height/2). Then resize the canvas by 2 in both directions and fill the empty space with copies of the tile:

http://koti.welho.com/thonkasa/Roinaa/hextile2.png

Now crop the image into 72x72 size (the exact placement of the cropping does not matter) and trim the corners so that it matches the standard hex:

http://koti.welho.com/thonkasa/Roinaa/hextile3.png

Now comes the moment of truth: what happens when we tile this hex? Let's see:

http://koti.welho.com/thonkasa/Roinaa/hextile4.png

Almost makes you believe in magic, doesn't it?

There are two important things to notice:

# The size and aspect ratio of the original texture are not preserved (unless it is already 54x36 pixels), but this can be a good thing. If you start with a square tile where the pattern is seen as from above, then it gets scaled down vertically, creating an impression of perspective (in Wesnoth, the terrain is seen from an angle, not directly above).
# The hex will contain the original pattern twice, that is, every unique feature in the original tile will appear in the hex tile two times. This may make the pattern to appear too densely repeating, but once you have the hex made, you can change something about one copy of each feature to counteract this effect (as long as you don't touch the edge pixels, the tile will stay seamless).

==Method 2: Rotation==

In the second method the original tile must be square, for it will be rotated 45 degrees. The rotation may prove useful in cases where the pattern have straight forms, like stripes or checkers. In that case the end result will generally look better if the lines go from SE to NW and SW to NE rather than from south to north and west to east. So if your pattern has these lines vertically and horizontally, you should use this method.

If the original tile is not square, scale it horizontally or vertically so that it is. Double the size of the canvas in both directions and fill the space with copies of the tile, creating a 2x2 pattern. Below is our example, a ceramic tile pattern:

http://koti.welho.com/thonkasa/Roinaa/hextile21.png

Next rotate the image 45 degrees and scale it so that the height becomes 144 and width 216 pixels:

http://koti.welho.com/thonkasa/Roinaa/hextile22.png

Then cut the hex shape out (anywhere where it doesn't include transparent pixels), and that's it (notice that the pattern appears only once in the hex):

http://koti.welho.com/thonkasa/Roinaa/hextile23.png

If you doubt that it can be this simple, let's see how it tiles:

http://koti.welho.com/thonkasa/Roinaa/hextile24.png

Which method to use depends on your requirements. If you need to preserve the original orientation, use method 1. If you need the rotation, or must have the pattern repeated only once per hex, use method 2 (both preserving orientation and having 1 repeat/hex is not possible with these methods). If the orientation and repeat number don't matter, the appropriate method depends on the desired size of the pattern: method 2 reproduces the pattern twice as big as method 1.

[[Category: Art Tutorials]]

Seamless Tiles Using Inkscape

2008-02-21T19:35:48Z

Sgt. Groovy:

In this tutrial you will learn how to use [http://www.inkscape.org Inkscape] to create seamlessly hexagonal tiles. The tiles are first made in vector graphics and then rendered into a raster of the right resolution. The method utilises the clone tile tool, and the cool thing about it is that you see a multiple hex area which updates immediately as you edit the original hex.

The method proceeds as follows:

First make a hexagon path with the Star Tool (set the number of corners to six and check the polygon button. Keep CTRL pressed to get the hex in correct angle). Set the width and hight of the hex path to 72px. This hex will be the background.

Create some object that fits inside the hex and group it with the hex (at this point it doesn't matter what it is, you just need a group and you can't group only one object). Give the group an ID which makes it easier to find in the XML editor.

Select the group and open the "Edit -> Clone -> Tile clone" dialog. Set the symmetry to P1 (the default) and the size to something like 5x5. Open the "shift tab" and set "shift X per column" to -25% and "shift Y per column" to -50%. Press create, and you have a hex grid made up of clones. The original hex is under the top left clone hex, so remove the clone to edit the original. Now any change you make on the original will be updated on the clones. See the image below for example, the selected hex is the original, others are clones (it is not meant to be any terrain, just random shapes for demonstration).

http://exong.net/wesnoth-attach/files/hexdemo_159.png

The tiling of the hexes is precice, but you can still see a one-pixel boundary between them caused by anti-aliasing bias (you can see this by zooming in, the boundary always stays one pixel wide). To get rid of this, CTRL-click (select within a group) the background hex path (the original), grab the bottom right scaling arrow and drag it out just enough to make the gap disappear. Now you are ready to edit.

One thing you must remember that you made the clone out of a ''group'', so you can't break the group without breaking the cloning. To select objects to edit, you must always use CTRL-click (or select in the XML editor). Adding objects in the group is a bit tricky too. Create the object first on top of the original, the drag-and-drop it into the the group in the XML editor. You can put objects on top of the hex border, but because of the z-ordering, you can only put them on top of the north, nortwest and southwest borders. This is not a broblem, though, because a copy of the object will always appear on the opposite border. If you have screen real estate to spare, you can keep two windows open, one with zoomed in view to the original, where you do the editing, and another with a zoomed out view of the grid in right size.

Once you are happy with the result you must render it to a Wesnoth terrain bitmap. Open "File -> Export bitmap" and export your image in 90dpi. This should make one image pixel unit equal one pixel in the export PNG. Open the exported image in raster program and crop it down to 72x72 pixels. The exact placement of the cropping frame doesn't matter (as long as it doesn't contain any of the empty parts), only the size. Now make transparent the pixels that would fall outside the stardard Wesnoth hex, and what you have left is a seamlessly tileable terrain hex.

Making the transition hexes for each border, and possibly some variations to break the repeating are best done in a raster program.

[[Category: Art Tutorials]]

Making portrait art in vector

2008-02-21T19:35:12Z

Sgt. Groovy:

This is a tutorial for making unit portraits for the Battle for Wesnoth in vector graphics. The Wesnoth portraits are made in what is called "the Lutesian style" after the comic artist Jason Lutes, who has donated many of the Loyalist portraits. The main reason for stylistic restrictions is to achieve aesthetic consistency.

The technical aspects of the Lutesian style can be summarised as follows:

* The colouring is made in "cell shading," that is, colours are applied in constant colour areas, known as "flats." Changes in tone (such as shadows and highlights) are also made the same way, instead of brush strokes or gradients.

* All the linework, instead of being black or having another colour of their own, is made in darker (or sometimes lighter) shade of the colour of the flat it is on top of.

* Adjacent flats don't need to be separated by outlines.

In addition to this, the final format of the portrait file should be a 205x205 pixel PNG with a fully black background.

Vector graphics programs can be easily used to produced this kind of images, because the flats can be easily crated and edited as paths filled with a solid colour. The linework, shadows and outlines are most conveniently made as semitransparent dark or light areas on top of the flats, because then they always follow the colour of the flats, if it is changed.

In this tutorial we use Inkscape as the program, but any vector drawing program worth its salt should have the same functionality. As a prequisite, you should be familiar with the basic concepts of vector drawing, how to make, edit and stroke paths, the fill rules (how to make transparent "holes" in paths), etc. Basically, reading the Inkscape manual and playing around with it for half an hour should be enough. You should also be able to draw and scan a pencil sketch four your portrait, or have other people's sketches available (remember to ask for permission).

Here's our example sketch, a young burgher:

http://koti.welho.com/thonkasa/Roinaa/burghersketch2.png

The blue lines show the edges of the shadows. It is convenient to make the shadow edges visible, because it is then easier to trace them (and scanners easily lose the lighter shades of pencilwork anyway), and by using a colour pencil, you make them separable from the actual linework. You can do the same thing with highlights, but use different colour than for shadows. The red arrow indicates the general direction of lighting.

The first thing to do is to trace out the flats and make outlines for those that are going to have them. As stated above, we are going to make the outlines as a separate, semitransparent path on top of the flat path, so that it gets its colour from it. Because lines differentiating two flats of different colours should have the colour of one or the other, but not both, it follows that the outer edge of the outline path must align exactly with the edge of the flat path. We will learn shortly how to accomplish this, but first read [[Z-order tricks|this tutorial]] about the z-order and the path tools.

Then it's time to get work on the image. First thing to do is to prepare the sketch. Since we must keep the sketch as the topmost object to see it as we draw, but also see the work beneath it, the background colour of the skecth must be changed into transparency. This can be done in GIMP, where the necessary tool can be found in "Layers -> Transparency -> Color to alpha." Once this is done, save it as PNG and import it into Inkscape. It will initially have some random X and Y values, so change them to 0, in the top toolbar. Open the document preferences and set the page size to match the selection. Rename the current layer to "sketch" and lock it, so that you don't accidentally move the sketch image, and it will be easier to select the objects under it. Create a new layer under the sketch layer and name it "flat." Now you are ready to trace the flats.

There are to ways to trace, with the freehand tool, where you move the cursor along the line and nodes are created on the fly, or the bezier tool, where you set the nodes explicitly. If you have a tablet and a very steady hand, you can manage with the freehand tool, but otherwise the bezier is better. The good thing about the bezier tool is that you end up with relatively few nodes, which keeps the file size smaller and makes later editing of the paths easier. Beginners usually make the beziers by first making an approximating polygon, and then coming back to round the corners. This is too inefficient, a faster and equally easy (after a bit of practice) way is the "tangential sweep." Follow the line and whenever there is a sharp corner, click on it without dragging. In between, click and drag tangentially to the line in few points along the curving line. You learn by practice where to place the sweep points. If there is a straight line between to corners, you don't need to put any nodes there.

The flats are best traced in back to front order, so that flats that are going to be the on the bottom are drawn first. It's usually good to base the z-order of the flats to the natural order of the real-life objects they represent, so that flats representing the foreground objects are on top of the flats representing background objects. There are times when you must break this order, but following it as a general rule helps to keep the drawing organised in your head.

In our image, we start with the sleeve at left. It's one of the "bottom objects," it's being covered by the torso, but doesn't itself cover anything. In the image below we see the sleeve traced with the bezier tool. Note that this is not the only, and not probably even the best, way to place the nodes. There actually isn't a "right" way to trace a shape, the only places where you ''must'' put a node are the sharp corners. You should try to get away with as few nodes as possible, though. You will also notice tha we didn't trace the line separating the sleeve and the torso. We will do it when we trace the flat for the torso, which comes on top of this one.

http://koti.welho.com/thonkasa/Roinaa/burgshot1.png

If we want the flat to have an outline, this is the time to do it, because it will then be directly on top of the flat it follows, and has the same relative z-placement as the flat.

The outline could be made with the calligraphy tool, but then we would be tracing the same lines twice. The flat could also be stroked, which would produce the outline automatically, but there are two problems. First, the stroke is always half outside and half inside the path, but we want the outer edge of our outline to follow the path exactly. Second, the stroke is always constant width, and this tends to result in dull and formalistic look, better suited to diagrams and such than artistic drawings. Even though strokes are not good as themselves, there is a handy tool "stroke to path," which makes a path out of a stroke so that the path follows the edges of the stroke, and we can use this to turn a stroke into a proper outline.

The "stroke to path" tool makes the original path disappear, so we will first do "Edit -> Copy" and "Edit -> Paste in place" for our flat. Then select the copy and disable fill and enable stroke. Set the stroke colour to black with 50% opacity (you might want to adjust this later, but these are good working settings at this point).

http://koti.welho.com/thonkasa/Roinaa/burgshot2.png

The width of the stroke should be about twice the intended mid-width of the outline. Then do "Path -> Stroke to path." Half of the outline still lies outside the path, and we will cut it away with a clip path, but wirst we will have to make the outline variable width. This is done easily by moving the outline path slightly, only so much that the edges of the underlying flat stay within it. Moving the outline makes the part that is inside the flat, that will remain after the intersection operation, to become thinner in some parts and thicker in other. As a general rule, the outline should be thinner on the lighted side of the object and thicker on the shadow side, so you should move the outline towards the light source.

http://koti.welho.com/thonkasa/Roinaa/burgshot3.png

When the inside part of the outline is the way you like, select the flat path and do "Edit -> Clone -> Clone". Select the clone and the outline path and do "Object -> Clip -> Set". What we did was using the clone of the flat as a clip path on the outline path, making the parts that are outside the path invisible. Because we use a clone, if we edit the nodes of the flat, the changes will automatically change the (visible) shape of the outline path correspondingly.

http://koti.welho.com/thonkasa/Roinaa/burgshot4.png

Since this method decides the outline width only by the angle of the outline, it doesn't always produce the desired outcome in all the places, but you can easily adjust the line width by moving the appropriate nodes of the ''inner edge'' of the outline path, inwards to thicken the line, outwards to thin it (the "sclupting" mode can become handy for this).

Proceed by this method for the flats in back-to-front order. Having the flats traced and outlined, my picture would look like this:

http://koti.welho.com/thonkasa/Roinaa/burgher1.png

Note that not all flats need an outline. Outlining usually adds a sense of weight and substance to objects, so thin, light and fluffy objects, such as hair, beard, moustache and the quill in my image, generally look better without.

Later it will be useful to have the silhouette of the whole figure in one path. Create a new layer titled "silhouette," select all the flat paths and do "Edit -> Copy." Switch to the "silhouette" layer and do "Edit -> Paste in place," followed by "Path -> Union." This will fuse all the paths into one, creating a silhouette. Turn the visibility of this layer off and lock it.

At this point it might be a good time to handle some special features, such as eyes. We could make them as flats on top of the face flat, but the pupils usually require adjustements to make them look just right, it's more convenient to make the eyes as holes in the face, and put the pupils behind it. If you make the pupils as whole ovals, only part of them is visible, but when you move them, the part always shows which should.

There are two ways to make a hole in a path. One is to make a path the shape of the hole on top of it and doing "Path -> Difference," but if you select a path and press shift while starting to make a new one, the new one will become a subpath to the selected one, and if it's inside it, it will appear as a hole (provided the fill rule for the path is even-odd, the default).

If you want the holes to have outlines as well, it will require a bit of trickery. Select the face flat and do "Edit -> Copy" and "Edit -> Paste in place." Then do "Path -> Break apart," which will make each of the subpaths into separate paths. Select the outermost path and delete it. What you are left with are paths that excactly cover the holes. Follow the outline making procedure, except in the end, instead of "Path -> Intersection," do "Path -> Difference" to clip away the ''inner'' half of the outline path.

For the eyes, you need white for the eyeballs (the white you see is the background colour). So make a eyeball colour flat (not usually pure white) that covers both of the eye holes, and drop it below the face flat. Then use the ellipse tool to make the iris and the pupil, group them, and drop the group below the face flat, but above the eyeball flat. When you have made one iris/pupil combo, you can copy it to get the other be exactly alike.

http://koti.welho.com/thonkasa/Roinaa/burgshot11.png

Now would be a good time to make the rest of the linework, that is the lines not outlining any flats. These lines are also to be made dark with partial transparency, so that they become a darker shade of the colour of the flat they are covering. There are several ways to make the linework. With short lines extending from the outlines, like the ones on the sleeves in my picture, can be made by making new nodes in appropriate places on the outline path and dragging them out. Thick lines can me treated as the flats by tracing the edges of the lines with the bezier tool. If you have a tablet, the calligraphy tool might be useful, but it takes some practice to use well. If the calligraphy lines overlap, you have to fuse them together by doing "Path -> Union" on them, or the partial transparency will produce darker shades on the overlapping parts.

Sometimes tracing linework is done more easily on a raster program, because you can use the eraser tool to trim the lines and correct mistakes immediately (especially with a tablet, when you can use the other end of the stylus as the eraser). The prosedure is as follows: Open the original sketch bitmap on a raster program. Create a new transparent layer on top and do the linework on this with black. Save only this layer as a PNG and import it into Inkscape. Set the X and Y properties of the bitmap to zeroes to make it align with the rest of the image. Select the bitmap and do "Path -> Trace bitmap" (the default options work well most of the time). Delete the bitmap and what you are left with is the linework done in paths. They are fully black so decrease the opacity and edit them further with the usual tools, if necessary.

Whatever method you use to make the linework, it's good to keep things organised and put it on its own layer.

http://koti.welho.com/thonkasa/Roinaa/burgher2.png

After the linework we can do the shadows. Create a new layer titled "shadow" on top of the others, and start tracing the shadow edges like you did with flats, but obviously, the shadows don't need outlines. You can make the shadow paths extend beyond the outlines of the figure, and cut them off in the end using the silhouette path you created earlier, using "Path -> Intersection." The shadows will also be partially transparent to show through the colour of the flat, but their own colour can well be something else than black. In my image I have used dark blue for the shadows, as in outdoors the light coming to the shadow parts comes from the sky and is bluer than the sunlight. You can try different colours for your shadows to see how it affects the general mood if the image.

http://koti.welho.com/thonkasa/Roinaa/burgher3.png

We still have some details to do, like the buttons on the jacket. Here I made the buttons with the ellipse too, using the same ellipse to create the shadow (by now you should be able to figure out how to make the crescent shape). If you need identical objects, you can always make copies of one, but if you make clones instead, and decide to change something about the design of the object later, you only need to change the original and the clones change as well.

http://koti.welho.com/thonkasa/Roinaa/burgshot12.png

To complete the image, some final touches are needed, like the specular spots in the eyes and some texture and highlights for the beard and hair (all done with semitransparent dark and light paths). Then the standard black background and some cropping and we're all done (I'll leave it as the exercise of the reader to figure out how to do the cropping).

http://koti.welho.com/thonkasa/Roinaa/burgherfinal1.png

[[Category: Art Tutorials]]

Shaped gradients with Gaussian blur

2008-02-21T19:33:58Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

The gradient tool in Inkscape is incredibly useful, but it has its limitations as well. For example, the radial gradient is limited to circular or oval in shape. To make gradients of other shapes, the Gaussian blur tool becomes handy. When you blur any object different in colour than its background, you are essentially creating a shaped gradient between those colours.

To make a "radial" gradient of any shape, that is, a gradient that changes from the edge of the path toward the centre, first create the path and set its fill colour to what you want the outer colour to be. Then make "Path -> Linked Offset", raise it to the top and set it to the inner colour (or usually somewhat more saturated one, to compensate the mixing when blurred). Use the node handle to make it smaller than the original path (1) and blur it to create the gradient (2). Because the inner path is linked to the outer one, any changes to the outer path will be updated in the shape of the inner one (though you may need to readjust the blur radius if you change the shape radically). To make multi-stop gradient, simply make several linked offsets of different colours and sizes (3). You will notice that the colour transition is smoother (rounder) than in normal gradients. If you need more linear gradient, you need to make several intermediary (in size and colour) linked offset paths.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial2.png

The same method is good for making any odd-shaped gradients. Just create a stepwise gradient with several paths and blur them to make the gradient smooth. In the image below there is a stepwise gradient created with the interpolate effect (this effect is especially handy for this, because it can interpolate the colour as well). In the lower image all but the outermost path have been blurred.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial3.png

One of the limitations of the gradient tool, when applied to strokes, is that if you want the gradient to change from the inner edge to the outer, you can only do this for oval-shaped paths. The blur, together with masking, can fix this problem as well. First create the base path, then make a linked offset path for it, raise it to the top and set its fill to none and the stroke to the outer colour (1). Create another linked offset, raise it to the top and set its fill to the inner colour, stroke to none and blur it enough to create the stroke gradient (2). Group this path with the one created in step 1. Create yet another linked offset path with no fill, and the same stroke width that the one in step 1, but with white colour (3) If you want to use a stoke style (dotted line etc.) set it for this path. Use the white stroked path as a mask for the group created earlier. Now you have a stroke with a gradient, and mostly it behaves like an ordinary stroke — edit the nodes of the base path, and the stroke changes with it. Even path operations work, as long as they produce only one path (if they produce several, the gradient stroke is only applied to the topmost one).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial4.png

[[Category: Art Tutorials]]

Variable-width strokes

2008-02-21T19:33:40Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

In SVG the strokes are always constant in width. This is good for diagrams, but makes drawings look too formal and dull. Variable width outlines can be made with the calligraphy tool, but it requires a tablet and a relatively steady hand. The drawback of this kind of lines also is that if you decide to change the shape of the underlying colour flat, you have to redraw the outline.

Howerer, with a little trickery it's possible to create variable width outline easily for a path of any shape that behaves like the normal strokes, that is, when you change the shape of the path, the shape of the outline is updated as well.

The process is outlined in the image below:

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial1.png

# Create your path with any tool you like. Set the stroke fill to none
# From "Path -> Linked Offset" create a path whose size (and shape) is linked to the original path, ''but do not change its size''. Raise this path to the top, set its fill colour to none and the stroke colour to the desired value (here it is black with reduced opacity, so that it will always show as a darker shade of the flat colour). Set the stroke width to about the sum of thinnest and thickest lines you desire.
# Move the outline path slightly to the direction you want the thinnest parts of the outline to be (here it is moved left and up). The original path should stay completely within the outline path.
# Make a clone of the original path and use that as a clip path on the outline path. This will make invisible the parts of the outline path that lie outside the original path.

Now you have an outline that will adjust to any changes in the size or shape of the original path, and whose thickness is determined by the path normal. You can use gradient fills for the line and line styles, like dotted line etc., but remember that only the inside half of the line is visible (normal strokes lie half outside and half inside of the path).

The downside of this method is that the line thickness and shape are determined automatically and are not always exactly what you would want. To make the outline more editable, select it and first do "Path -> Object to Path" and then "Path -> Stroke to Path". The inner egde of the outline path is now a subpath that you can edit as you whish, but the line isn't linked to the original path anymore, and so doesn't keep up with any of its major changes. It's still clipped by the clone of the original path, so it will adjust to small tweakings.

[[Category: Art Tutorials]]

Z-order tricks

2008-02-21T19:33:19Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

In a vector image, every object (path, rectangle, piece of text, etc) has a position in the x-y plane, but it also has a position in the depth dimension. If you think the objects as paper cut-outs, any time two of them overlap, one must be on top of the other. Similarily, each object in a vector image occupy a slot in the z-order, which dictates which object appears on top of which (the difference to paper cut-outs is that they retain their z-position even when they are not overlapping in x-y space). You can always rearrange the z-order of your objects, within certain limits. For example, when you group objects, they retain their relative z-order to each other, but no object outside the group can occupy a z-slot between the ingroup objects.

You can use the z-ordering to you advantage, when you want areas of different colour to appear next to each other without a gap in between. As with paper cut-outs, the simplest way to do this is just to make one area to expand a bit under the other. This has the benefit that you can later adjust the positions of the paths without a gap appearing.

This approach is not possible in all situations, however. One common situation is where you have objects with a circular z-order: Object A covering part of object B, which covers part of object C, which in turn covers part of object A. In a vector image, the z-order is always linear, and to achieve the appearance of circular z-order, we need some trickery, the tools for wich are found in the Path menu.

In the image below we have an illustration of the problem. We want three ellipses of different colours stacked circularily as in top right, but with a linear z-order, the top left is best we can do. We can't put the red ellipse under the yellow one without putting it also below the green one, because that is on top of the yellow.

http://koti.welho.com/thonkasa/Roinaa/burgshot13.png

In order to achieve the desired effect, we need to cut off a piece out of the red ellipse that corresponds to the part that would be under the yellow one. To do this, you will first select the yellow ellipse, do "Edit -> Copy" and "Edit -> Paste in place." This will result in a copy of the yellow ellipse in exactly the same place. Then select both the copy and the the red one, and do "Path -> Difference", which will remove from the red ellipse the area overlapping the yellow one. In the bottom image is a copy of of the ellipses, with the red one moved outwards to show how it's shape is changed. This also demonstrates the drawback of the method: After you have matched the edges, you can't move the paths in relation to each other without breaking the alignement.

There is another, a bit more complex way to do this operation, which will allow the paths to be edited or moved without breaking the alignment. First select the red path and make "Path -> Linked Offset", creating a copy of that will change when the original is changed. Raise it to the top and change its fill colour to white. Do the same for the yellow path, but change its colour to black. Select both linked offsets and group them. Select both the group and the red path, and do "Object -> Mask -> Set." After this operation the red path is masked with its own linked copy in white, combined a linked copy of the yellow path in black. The result is that the red path is always invisible where it overlaps the yellow one. You can edit the nodes of either path, and the mask is updated as well. You should note, however, that when a masked object is transformed (moved, scaled, etc.), the transformation is also applied to the mask, which is not desireable in our case. So if you want to do transformations on the red path, disable the mask first, do the transformation and reapply the mask (you can move it in node editing mode without this hassle, though, by simply selecting all the nodes).

[http://tavmjong.free.fr/INKSCAPE/MANUAL/html/Paths-Combining.html Here] is a summary of the path operations and how they work. They can be incredibly powerful once you learn how to use them creatively.

Let's now look at one more, a bit trickier case. Sometimes you can have a circular z-order between ''two'' objects. A good example is Saturn and it's rings, coming from behind the planet and then going over it. We see it in the image below, bottom right is our goal, and top left are the original paths, created with the ellipse tool (now you should already know how to cut the "hole" in the blue one).

http://koti.welho.com/thonkasa/Roinaa/burgshot5.png

We know that we can use a copy of the red path to cut off the part of the blue one that should be eclipsed, but how do we do that without cutting off the part that is supposed to eclipse the red path? First make "Edit -> copy" and "Edit -> Paste in place" for the red path and select this copy and the blue path. But then, instead of the "Path -> Difference" do "Path -> Division", which will cut the blue path into several paths along the red path's edge, but leave them all in place (top right). One of the parts corresponds to the section that should be covered by the red path (bottom left). Select this path and delete it. The remaining paths now make up the part that should be visible, but it is still divided into separate paths. Select them all and do "Path -> Union," which will fuse them back into one path again.

[[Category: Art Tutorials]]

Designing weapons and armour

2008-02-21T19:32:59Z

Sgt. Groovy:

by Zookeeper

Often one sees that someone draws these things looking all silly. Slight exaggeration of weaponry and such is of course fine in a fantasy setting, but it's not uncommon at all to go over the board and draw something that couldn't function in reality at all. This mostly applies to portraits and other realistic imagery, not really unit graphics as such.

Now, my point is simply: you can make weapons and armour be flashy and cool while keeping them very realistic, and a real-world reference also helps you to get things functional without having to think what kind of sword handle would work and what wouldn't, for example (and, no offense to artists drawing fantasy stuff, but that usually goes a bit awry), and that there are plenty of references available. So even if you don't feel the need for the weapons you draw to look realistic, having some references still helps to actually make it look good (in the same way as having a reference when drawing a person makes it much easier to get it look right and good).

For this purpose, I thought I'd go through some related sites I know of and post some links here. These are armouries, general information/review sites, etc, which contain good pictures of historical arms and armour. Mostly (if not all) are only european stuff though (which sort of fits with at least the mainline Wesnoth setting), but that's all I had available right now.

* http://www.armouronline.com/ - the full body armour section is good, weapons are generally crappy here.
* http://www.myarmoury.com/ - hard to navigate, but contains a lot of good pictures, if you take the time to find them. The "reviews" page is a good place to start from.
* http://www.lutel.cz/index_.php?en=1 - quite a number of all sorts of weapons in the catalog (especially the polearms and such look nice).
* http://www.darksword-armory.com/ - a collection of often historically accurate weaponry
* http://www.by-the-sword.com/ - various things
* http://www.siue.edu/COSTUMES/history.html - a site with lot of pictures of clothing from different eras and areas, from 19 century engravings. The scans could be of higher quality, but you can find plenty of stuff for reference/inspiration for fantasy clothing. Many of the same pictures can also be found in a book ''A Pictorial History of Costume'' (Pepin Press, no author).

Feel free to post your additions (especially to the type of items not covered on these sites), too, since we don't have much here yet.

[[Category: Art Tutorials]]

Scaling Digital Images

2008-02-21T19:32:13Z

Sgt. Groovy:

Scaling of digital images is subject to the same problems that come with and without antialiasing. Rasterized digital images are composed of a mosaic of little square pixels. When you resize an image, a computer will create a new mosaic of pixels based on the previous one. There are several methods that a computer can use to do this, which have slightly different results.

Interpolation is the process of doing this resizing, and of choosing what colors should go into the new pixel mosaic. When scaling an image to larger size, there will be more pixels present; there will be cases where there are now two pixels where there was one before. When scaling an image to a smaller size, there will now be pixels saddled with the task of representing the same colors that more than one pixel did before.

Nearest Neighbor interpolation, sometimes called a lack of interpolation altogether, decides which color the new pixel will be based on a simple criterion. Imagine that instead of resizing the mosaic, that we were instead remaking the same thing at the exact same size, with many more (smaller) tiles - this would be analogous to scaling an image to a larger size. One way to do this would be to simply use the exact same color as whatever previous tile was closest to the new one you're putting in. In fact, you could simply break the old tiles into little pieces where they lay, and you'd get the same effect.

The problems with this are simple - if you are scaling up, the tiles don't blend in color any more than they used to, which wastes the presence of all those new tiles. The second problem is that if you are scaling to a smaller size, with less tiles, you will run into situations where two are being replaced by one. In theory, if you had a blue tile next to a red tile, it might make sense to replace the both of them by a single purple tile; however, you are forced by the "Nearest Neighbor" method to choose one of those two and remove the other entirely.

Bicubic interpolation, and Bilinear interpolation, do the blending I just talked about. There are many other methods which work something like these two, and when given a choice between them, always choose to use Bicubic Interpolation over Bilinear interpolation. It's simply better. It is very slightly slower, but this hasn't been an issue on personal computers for quite some time, and the continued availability of it is just a legacy of a time when computers were much worse than they are now.

This image illustrates the bad effects of downscaling with Nearest Neighbor filtering:

http://exong.net/wesnoth-attach/files/konradinterpolationcompare_656.png

Side note: One of the problems with digital scaling is trying to scale to a larger size. A computer can scale any image to a smaller size without any problems, but to scale something to a larger size is difficult, because you're trying to fill in details that simply do not exist. Nearest neighbor filtering will produce hard, rectangular edging, bicubic will blur the space between the pixels. In certain rare instances, such as old-fashioned sprite art, special algorithms that follow the edges of similarly colored pixels can be used to better effect than either bicubic or nearest neighbor interpolation. These are often used in nintendo emulators. It should be noted, however, that these would be useless for wesnoth, because of how our sprite art is drawn.

===Which Scaling to Use:===
based on the type of art being scaled.

For sprite art, it is generally best to set your scaling method to "nearest neighbor." After scaling your image, you '''absolutely must''' retouch any misplaced pixels by hand, and I guarantee they will be present. Sprite art is very difficult to scale, because of the procedural nature of its construction. The smaller a piece of sprite art gets, the more important a single pixel and the arrangement of pixels becomes; thus it follows that any displacement of these can destroy the effect they had before.

For basically all other art, use the "Bicubic Interpolation." Be this photos, scanned drawings, cell-shaded art, anything.

===How do I set this?===
(Note that this describes the global setting. If, in any scale dialog box, there is another menu with the same thing in it, that sets the same thing, but only for that individual operation.)

In the Gimp, you should find a panel in the preferences that looks a bit like this. The section outlined in red is the place you change this:
http://exong.net/wesnoth-attach/files/scaling_363.png

In photoshop, the following picture illustrates the same thing:
http://exong.net/wesnoth-attach/files/interpolationphotoshop_513.png

[[Category: Art Tutorials]]

Inking With Pencils

2008-02-21T19:31:17Z

Sgt. Groovy:

====Foreword by Jetryl:====
This information was given me by a contributor named Jason Lutes, a professional artist who contributed a number of very well-done portraits for this game. I asked him a question about his workflow, and he kindly responded.

"Inking," in cell-shaded art jargon, is a method of making pencil lines scanned into a computer more solid. When you draw on paper with a pencil, it will only stick to the fibers of the paper that jut out higher than the other fibers, this means that when you look closely at the line, it will have holes in it, like a chalk line drawn on rough asphalt. This is very apparent with pencil drawings that have been scanned into a computer. Professional work does not have these holes in the lines, and traditionally, professionals have achieved this by tracing over their pencil lines with ink (hence the name). Ink soaks into the fiber and gets pulled into even the low-lying fibers, leaving a solid, uniform, black line. There are other methods of doing this, however, thanks to the advent of computers; one of these is described below:

===How to make pencil lines look like inked lines:===
The drawings I've posted so far are actually scanned in from drawings done with a mechanical pencil (HB lead) on vellum (heavy tracing paper) -- they're not even inked. The pencil sketches are very tight, but I have a little Photoshop routine I run to get them looking smoother.

I scan in the drawings at 300 dpi, then I adjust the contrast until the lines darken but don't thicken up, and the background is pure white. Then, I use the magic wand to select all the line work (tolerance around 60%, constrain unchecked -- but tolerance adjusted depending on how it looks when I zoom in close to see what I've grabbed), and fill it solid black. Then I Select -> Inverse and fill the rest of the image with solid white, so the only two shades on the image are 100% black and 100% white.

I usually run Filter -> Noise -> Despeckle, then the trick move is Filter -> Noise -> Median, set to 1 or 2 pixels depending on how the preview looks. What this does is check every pixel on the image against every adjacent pixel (within the pixel radius you've entered), and smoothes out the jagged parts. It's a pretty good trick.

I think you need to start with a scanned image of decent resolution for it to look good; the finals that I've posted are probably only one-third the size of the original scanned version, so the shrinking helps to make things look good too.

I do encourage people to try inking by hand. It takes some practice, but with continued effort you will get to the point where you're happy with your linework. When I ink, I use a Rotring ArtPen, with a refillable cartridge of non-clogging, waterproof India ink. You don't have to go to that length though -- find a pen that feels good, and find a smooth drawing surface. Professional cartoonists and illustrators who have a very clean look to their art often use a paper called "bristol board," of which there are several types. You want the kind with "plate" finish, or, if that can't be found, "smooth."

For scanning images into a computer, though, as I said, you can probably get smooth lines with a lot less effort, if you try to follow the process I've outlined above.

[[Category: Art Tutorials]]

Art Supplies

2008-02-21T19:30:56Z

Sgt. Groovy:

What "physical art supplies" (not counting computer programs, although I make an addendum about them) you need to draw the way that professional artists like [http://www.machall.com/ Ian McConville], or [http://www.applegeeks.com/ Mohammad F. Haque], or certain artists who have contributed to wesnoth, do.

Not much, really. You need a good pencil, decent paper, and a few good erasers. That's it. One thing I should drive home is that buying great art supplies; or in fact, buying things specially set aside as art supplies, will not make you better at art any more than buying a better football will make you a good football player. You simply need something that is functional - to extend the football metaphor, it would be worth getting a better ball if yours had a leak and was always a bit flat, or if yours had some funny lump in it; but any brand new ball from a local store would be fine. With these things, there is a point of extremely diminishing returns. Unless you are ''extremely'' good,'''if you spend more than $25 on art supplies, you are probably wasting your money.'''

====The Pencil====
A regular graphite pencil, or a good mechanical pencil, will do the job.

I (Jetryl) do all of my drawings with a mechanical pencil, bought at a local store. I've grown quite fond of a specific make - the Pentel e-Sharp series with its "Lead-Maximizer" technology, and have grown so because that specific brand has a slightly unusual tip, which holds the lead more tightly throughout the life of a single stick of graphite. Generally, towards the end of a stick of graphite in a mechanical pencil (especially the cheaper "Bic" variety), the stick will no longer be gripped by the lead advancer inside the pencil. This allows it to rotate freely, and also slide in and out of the pencil (potentially falling out, if you held the pencil in the open air, pointing straight down). This causes a loss of certain fine control, and sometimes causes accidental strokes when the pencil is close to the page and the lead falls down and onto the surface of the page. A good mechanical pencil like mine solves this problem.

Certain professional artists will get these things titled "lead/graphite holders" - not devices that advance the lead for you, but ones whose head simply grips a stick of lead like a vise and have a pencil-like outer shell for you to grasp and draw with. These solve the same problem.

Amongst mechanical pencils, .5mm lead pencils are probably the best bet; they offer better control of details than .7mm pencils. .9mm, .3mm, and other sized pencils can be found for purchase, though they are considerably harder to find than the .5mm and .7mm pencils available at your local corner store. When I need something sharper than the .5mm width point, I take a very sharp piece of metal (a razor blade works well), and hone the tip of the graphite stick down for maximum precision.

[http://en.wikipedia.org/wiki/Pencil Traditional pencils] made with a shaft of graphite wrapped in wood come in many varieties of hardness, and can be used to shade in areas with respectively varying control over the level of shading (typically, softer lead gives darker marks, harder lead, lighter marks).

Total cost of a few decent pencils: $3-5

====The Paper====
Paper, like ice for a skater, can actually mess with your work if it's too rough. You (like myself) can get away with using nothing but mere printer paper, rather than buying fancy art paper. Just be warned that while some works beautifully, certain varieties aren't so reliable. You want to get paper that is very bright (probably about 95 or so, at least by HammerMill's rating system), and paper that has an extremely smooth surface, which you can only judge by trying it. An excellent way to do trial and error is to swipe a few sheets from any copiers you encounter; usually they'll each be stocked with a different brand, and they'll have spare, labeled, unopened reams inside or near them, so you can identify the brand.

Paper is composed of a bunch of fibers - on a microscopic level, these look vaguely like a bunch of chunks of hay or grass mashed together. Or perhaps like a rough, hilly topography. When you draw with a pencil, the lines you make generally only show up on the peaks of those "hills", on the fibers that stick up higher than their peers. This means that there are little gaps in your lines. On really bad paper, these can become quite apparent, and can be so bad that, like grooves in a record guiding a needle, they cause your pencil to move in directions you don't want it to. That can foul up finer details in a drawing. If the paper is causing these problems, it's not good enough. I suggest buying the best ream (500 sheet pack) of laser printer paper you can find in your local department store. This should cost you no more than about $6-10.

Professional artists sometimes buy [http://en.wikipedia.org/wiki/Cotton_paper quality paper] called "bristol board" or "vellum" for their illustrations (such as those mentioned at the start of this tutorial). It's only worth the money when you get really, really good at illustration; I myself am not good enough to capitalize on the difference, so I don't buy it.

Total cost for a ream of good paper: about $10

====The Erasers====
Your best friend.

A [http://en.wikipedia.org/wiki/Kneaded_eraser kneaded eraser] is a must. This is your primary tool. These are versatile things, and you may find them very strange if you're unfamiliar with them. They can be deformed like putty, often useful for making a point with which to erase fine details (a task they excel at), and also do not leave "shavings". They pull the graphite into their own fibrous mass, removing it from the paper as you rub the eraser against it. Because of this, they will over extended use become filled with graphite, and thus useless for pulling more out - it's at that point that you throw them away.

[http://en.wikipedia.org/wiki/Eraser Other erasers] can also be very useful, including the white ones attached to most mechanical pencils. I use an "art gum" eraser for any "nuke and pave" tasks, wherein I want to erase a section of a drawing back to as close to pure white paper as I can get. It leaves a lot of shavings, and is a unwieldy tool for fine details, but does that single job very nicely. White vinyl erasers, and many other various brands/varieties can be used for similar purposes. Kneaded erasers, especially after some use, don't work especially well for nuking something down to white, and thus a few other erasers to complement them works well. The non-pink eraser at the end of my pencil works rather similarly.

You should avoid the ubiquitous "pink erasers" often seen in school classrooms, and also avoid anything harsh enough to tear up the page (like certain erasers I've seen which are designed to erase ''pen'' markings). The classic pink erasers don't do so well at lifting the graphite off the page, and sometimes have a detrimental effect of grinding it under the fibers, at which point you can no longer erase it without removing a layer of the paper, which is often unfeasible or destructive. They also have a tendency to dry out with age, at which point they're largely unable to erase, and will cause damage to your drawing if you try to use them.

Total cost for a batch of three erasers: about $10

====The Computer====
If you want to color drawings on a computer, which I don't suggest trying until you can do really good pencil lineart, then you will need a few things. Any computer of about 500mhz, with a copy of the Gimp or Photoshop will do well. In 2006, these are trivial to acquire - people throw these kinds of machines away, and you might be able to get one for free, if you don't have one already. That underscores the real point, here, which is that the machine you're reading this on is more than likely ''grossly'' overqualified for doing this kind of work. You'll also need a [http://en.wikipedia.org/wiki/Image_scanner scanner], which will set you back about $100 for a decent one, and I very strongly suggest getting a [http://en.wikipedia.org/wiki/Graphics_tablet graphics tablet], which costs about $100 as well. Wacom sells the very nice Graphire tablets, of which the 6x8 size is perfectly adequate for anyone's needs ([http://www.goodbrush.com/ Craig Mullins] has done his work with tools worse than what I listed above, as have many other artists).

For drawing in a cell-shaded style like this, I do not suggest coloring with anything other than a computer. People used to use things called [http://en.wikipedia.org/wiki/Screentone screentones] to color cell-shading art in the past, but computers are better than them in practically any way, especially efficiency of work, and being able to correct mistakes..

Total cost for a great scanner and tablet: less than $200

Total cost for a suitable computer: free, you probably have one.

Total cost for [[Art_Programs|drawing software]]: free

[[Category: Art Tutorials]]

How to Anti-Alias Sprite Art

2008-02-21T19:30:27Z

Sgt. Groovy:

Anti-aliasing is a technique designed to counter the inherent "grain" of computer displays, a grain that results from their cartesian nature. Despite early experiments with radically different display technologies, the computing industry settled on this display system of rectangularly arrayed pixels, and to effectively use it, you must overcome one of it's greatest inherent limits. Anti-aliasing is designed to do just that.

The problem is this, when rendering a line, distances between drawn elements on a screen are not infinitesimal. You can only make a change to your image at fixed distances, specifically the true dpi resolution of your current display setting. As an aside, this is not to be confused with the display setting recognized by the operating system - macintosh screens were only 72dpi (dots per inch, lengthwise) on the original macintosh computers, though they have inherited that setting in name, it became meaningless with the advent of the first color screens. Likewise, though they pretend that they do, computers running windows very rarely have a monitor set to exactly 96dpi. Something near or above those is not uncommon these days, though.

To render a solid line, you cannot simply place pixels (picture elements) in the target color, as you might when drawing with a pen on paper, because the granularity of them on the screen is so big that you can see the breaks between the rows. The result of drawing a solid black line on a white background, without antialiasing, will look like this:

http://exong.net/wesnoth-attach/files/aliased_line_144.png

Notice the rather garish and sudden shifts between pixel rows? These have many names, including the "jaggies", and are an unnecessary evil.

This is how much smoother the line looks if you anti-alias it:

http://exong.net/wesnoth-attach/files/anti_aliased_line_634.png

Anti-aliasing copes with this by doing what any digital camera would do, in trying to decide what color a target pixel would be. Imagine the target pixel as a rectangular area of different colors that you have to represent by a single color. Imagine doing the same thing to a photograph. How would you go about this? By taking an average, that's how. If you were doing this to a photograph of the empty sky, you might pick a light blue color. When you are dealing with a line, you are dealing with a geometric object with width. You're not really drawing a line, you're drawing a solid rectangle. It's only in math that lines exist with no width - the lines you draw exist to pretend they are the "edges", the infinitely small exact points at which one thing starts and another ends.

In doing pixel and sprite art, you assume that your lines are usually 1 pixel in width. To anti-alias, you, in your imagination, superimpose the line on a grid representing the real pixels you're drawing on to. If the real, geometric line that you are trying to represent covers up a pixel completely, that pixel is the color of the line. If the real, geometric line doesn't touch the pixel, the pixel gets no color from the line. If the line covers it up partly, which is most common, the final color gets mixed towards the color of the line by exact fraction of how much the line covers the pixel.

http://exong.net/wesnoth-attach/files/anti_aliasing_212.png

Covering it all the way gives a full 1/1, and it goes all the way to the line's color. Covering 1/2 of the way moves it half of the way to the color - if we have a black line on a white background, halfway between those two colors is a medium gray. Covering just a tiny corner of the pixel with the line will mix in just a little bit of the line's color into the pixel - for the example we've been using this whole time, it'd end up a very light grey.

===Applications to sprite art:===

Anti-aliasing is how you make anything that is not a vertical or horizontal edge look good. You can get away with not doing anti-aliasing on many parts of your images, but if you do it some of the time, on the parts where it is most needed, it will make the image look much better. Anti-aliasing is often used on things like staves, or swords, when they hang at a slight angle rather than straight up or straight down.

* One really nice trick is that mixing colors by transparency works exactly the same way as the color mixing I described above - in fact, it's the exact same thing. If you anti-alias the edges of your units, where they interface with the terrain, they will look to have edges that follow the actual shape you intended, rather than being forced into a cartesian grid. You can create pixels along the edges of your units that are the same color as the edge, but with variable transparency. We can do this because, though we don't know what it's going to be mixed with, we do know how much of the unit's color to mix with the background, whatever that background is. It follows the same rules as those for the line I showed above. That "how much of the line's color" is the level of opacity. To do this, you will need a program which can actually work with transparency values, like Adobe Photoshop, or the free Gnu Image Manipulation Program (GIMP).

* One advanced way to have naturally anti-aliased lines is to draw your sprites with a paintbrush instead of the pencil tool. This is, however, a very difficult technique, and not something to be attempted unless you are a master at sprite art.

== See Also ==

* [[Create]]
* [[Create Art]]

[[Category: Art Tutorials]]

Scanning with camera

2008-02-21T19:29:20Z

Sgt. Groovy:

All art to be used in Battle for Wesnoth must be in digital form. Art made in real-life techniques must be transferred to digital format, and most often this is done with a scanner, but a digital camera can also be used. In this tutorial we go through the basics for getting good digital photos of paintings and drawings.

The good reasons for using a digital camera instead of a scanner include:

* You don't have a scanner. If you only intend to use the digitised image as a sketch to draw over, the quality requirements are much lower and often a photograph will do just fine. Even then, though, there's no reason why the quality should be any lower than what you can accomplish.

* The surface properties, such as crumpy oil paint or metal finish don't work well on a scanner.

* The image is too big to fit into the scanner. If the scanner lid is detachable, larger images can be scanned in pieces, but then you are left with the problem of fitting the pieces seamlessly together. With a camera you can shoot an image of any size with one shot.

To get a good shot of a painting or drawing there are two factors to consider: how to set the image in front of the camera, and how to illuminate it.

==Camera setup==

To avoid perspective distortion on the shape of the image, the view angle of the camera should be perpendicular to the image and pointing at the centrepoint of it. Also, to avoid the "fisheye" distortion, where the edges of the image are bulging out, the camera needs to be placed far enough, at least at a distance of several times the width of the image. Usually this is so far that a horizontal setup (that is, the line of sight is horizontal, and the image vertical) is more convenient. Any vertical surface is good for suspending the image, but if you use fixed light sources, you need something moveable. A table turned on its side will do nicely, if you have nothing else.

The camera needs to be fixed to get as sharp image as possibe, so use a tripod if you have one, or at least put the camera on a table, a box, or other stable surface. Zoom in enough so that the image fills most of the view, but it's generally best to do the final cropping on the computer. Set the camera on highest resolution and best image quality (least compression). Also, when digitising colour images, set the white balance to match the type of lighting you are using.

Before starting with lighting, you need to remove any glass covering the image, the surface of the glass is way too reflective (even with the "non-reflecting" types).

==Lighting==

The first thing to do to set up the proper lighting is to turn off the flash. The worst way to light the image is to make the light come from the same direction where the camera is. While the view angle should be perpendicular to the image, the best lighting angle is roughly 45 degrees. The reason is, that you want only the diffuse reflection of the light to hit the camera.

Another way to make the reflected light more diffuse is to make the incoming light diffuse. This can be accomplished by increasing the light-emitting surface of the light source. Thre are two ways to do this, the first is to pass the light through a diffuse filter. Anything translucent and milky will do such as a thin paper (baking paper), smoky glass or a muslin cloth. Photography shops sell special filter film for the purpose. Notice that the filter always reduces the light intensity. Another way is to use diffuse reflector, any matte white surface will do, but quite a lot of seemingly white materials actually absorb quite a lot of the light. Specifically good material for a diffuse filter is a plain sheet of polystyrene foam (styrofoam, styrox). It is both highly reflective amd dispersive, and as a ridig yet light material it is easy to suspend (often just prodded on a stick). The reflectivity can be further increased by painting the surface with a silverish metal paint, though it must be water soluble so it doesn't eat away the polystyrene. Aluminium foil can also be glued (with water soluble glue) on it, rubbed in so that it adopts the surface roughness.

Also, in order to get a uniform illumination of the image, the lighting needs to be symmetrical in reference to the view line of the camera. This means that you need, at minimum, two roughly identical light sources, positioned symmetrically on both sides of the camera. We are talking about radial symmetry here, so if you use more than two lights, they need to be positioned on regulal intervals on a circle that has the camera view line as its axis (but usually two lights do just fine).

The image below shows an ideal setup for two light sources (camera is blue, the image red, and the light sources yellow).

http://koti.welho.com/thonkasa/Roinaa/camera.png

Some of you might have gotten discouraged by all the talk about making reflectors and stuff, but things can be done as well with natural light. Windows are natural large area sources of diffuse light and the above illustrated setup can be achieved in a room with two windows on the same wall. Set the image on the centre of the room, facing windows, and put the camera next to the wall between them. There's also plenty of light outside, but direct sunlight is not good, you should only use light reflected from the sky or clouds. The best weather for photographing images ouside is a day when there is a cloud cover just thick enough to make shadows disappear.

==Testing your setup==

Once you have everything set up, you can run a test by taking a picture of a white sheet of paper the size of your image. Upload the picture into your computer and open it in a image processing program, such as GIMP or Photoshop. Using whatever tool available for making rectangles, you should be able to draw one that pretty much equals the shape of the paper, if your setup is correct. If the edges of the paper seem to bulge outwards, your camera is too close. If the shape of the paper looks like a trapezoid (opposite edges are not the same length), the angle between the view line and the image is not 90 degrees and/or the camera is not pointing at the centre.

Next you must use the colour picker (pipette) tool to check that your lighting is uniform and that the exposure and white balance match your lighting. Pick a value at the centre of the paper and close to each corner. The values should be roughly the same, and if not, your lighting is not symmetric enough. The values should also indicate no colour, that is the red, green and blue values should be roughly equal. If not, the white balance is not set right. The values should also be as close to full white (255,255,255 in RGB) as possible, but to avoid overexposure, slightily below it (because overexposure is capped to 255,255,255, you can't tell it from values that really are full white). If the values fall significantly short of 255, you need a higher exposure setting for the camera, or more light.

It is possible that no single setting of exposure will give you enough resolution in luminosity (dynamic range) in both the lighter and darker parts of the image. In these cases, the technique for extending dynamic range presented in [[Extending dynamic range|another tutorial]] will be applicable here as well. Obviously, any other postprocessing photo enhancement techniques can also be applied, but usually they also degrade the image quality in some way (e.g. intoduce posterisation or noise), so it's always worth the effort to make the image as good as possible already in the camera.

[[Category: Art Tutorials]]

Extending dynamic range

2008-02-21T19:28:56Z

Sgt. Groovy:

In the area of imaging, dynamic range means the imaging hardware's ability to tell differences in luminosity. Due to the non-linear nature of the human photoreception mechanism, the dynamic range of human perception is very large, and it's difficult to capture sceneries with high range of luminosities. A typical example is a scenery against a background of bright clouds. If you set the camera exposure to a setting suitable to the brightness of the foreground, the luminosity of the sky goes outside the dynamic range of the camera and most of the sky pixels are clipped to white, resulting in ugly flat "burnt through" white sky. On the other hand, if you set the exposure low enough to bring out the details of the clouds, the foreground luminosity drops below the lower end of the dynamic range, most of the pixels get clipped to black, and you are left with nothing but a dark silhouette.

There is a trick to extend the dynamic range of a camera: by taking shots of the same scenery in different exposures we can capture the details of both highligth and shadow parts of the scenery, which can then be combined digitally in one image. The picture below is an example of this technique. The leftmost image was taken with an exposure setting suitable for the foreground, but it got the clouds blended into a white haze. The middle image was taken with exposure low enough to make the cloud structure visible, but now the foreground is too dark. The rightmost image is a combination of the two others. The foreground pixels are mostly from the high-exposure image, while the sky is a blend of the two. The result presents the details of various luminosities much better than either of the other images and is closer to what human eye would see. The good thing about this technique is that no extra equipment nor fancy sofware is needed, the pictures were taken with a regular cheap digital camera (a film camera would also do, as long as you can adjust the exposure) and the postprocessing was conducted with GIMP, using a simple plugin downloaded from the plugin registry.

http://koti.welho.com/thonkasa/Roinaa/dynrange.jpg

Scanners also have a limited dynamic range, and similar problems can occur when scanning images that depend on very small differences in luminosity for detail. Pencil drawings typically fall into this category. In order to read the detail in the darkest parts, plenty of light is needed to penetrate the pigment to reflect back from the paper, but this much light washes away the faintest pencilwork. The pictures below demonstrate the problem, scanned at different exposures. The darker one has all the lighter shades and linework present, but has the darker parts lacking in contrast and detail and the whole picture is generally too dark. The lighter one brings the detail of the darkest parts out better, but the rest of the picture is totally washed out.

http://koti.welho.com/thonkasa/Roinaa/stoat1.jpg

http://koti.welho.com/thonkasa/Roinaa/stoat2.jpg

The solution to the problem is exactly the same as with the photographs, combine the different exposures into one image.

Some notes about scanning the different versions: most scanner sofware offer "image enhancement" options by changing the gamma, brightness and contrast of the image. These, however, usually only manipulate the data ''after'' it has been scanned, and don't really add new detail to the image that we need. What you must change is the actual brightness of the scanning head light source (or the sensitivity of the CCD, whatever option is available for your scanner). In Xsane, these options are found in the "Standard options" window. Also, when using Xsane, when you change these settings, the preview window is not updated until you press the "Acquire preview" button. You must scan exactly the same area for each exposure, so to make sure it doesn't change between previews, uncheck the "Preselect scanarea" button in the "Preferences -> Xsane setup -> Enhancement" menu.

Open GIMP or some other image manipulation program and put the exposures in the same image as separate layers. Simplest way to combine the images is to change the opacity of the upper layer to get a uniform blend. Most to the times, however, you want to mix the images selectively, by making some parts closer to the low exposure image, and other parts closer to the lighter one. This can be achieved by setting a mask to the upper layer. Paint the mask with 50% grey and you have a equal mix of the two layers. Now paint with black to the mask in parts where you want your image to be closer to the lower layer, and with white where you want it to be closer to the upper one. This method is labourous, but it offers you complete control of the result.

The mask can be also produced automatically. In GIMP, there are two plugins for that. They don't ship with the standard distribution, but you can download them from the GIMP plugin registry. The first one is called [http://www.trsqr.net/photokit/dre.html Dynamic range extender] and it presumes two exposures, the darker one on top. Uncheck the "merge layers", because you might want to try different opacities for the upper layer. The plugin creates a mask for the darker layer, which makes the the lower layer selectively visible.

The second one is called [http://registry.gimp.org/plugin?id=9520 HDR Tone Mapping], and it presumes three different exposures. The order persumably doesn't matter, and it produces a mask for the two upper layers (though in all my trials the mask of the upmost layer was almost totally black, thus not contributing significantly to the combined image). NOTE: this plugin presumes an RGB image, so if you scanned your images in greyscale, you must change the image into RGB before using this plugin (you can change it back to greyscale afterwards). If you try to use it on a greyscale image, the plugin just hangs indefinitely.

With both plugins, it's worth to try to change the masked layer's opacity after you have run the plugin. I have found that it often results in a better image.

Here's the result of the first plugin applied to the exposures above, with the masked layer's opacity set to 70%:

http://koti.welho.com/thonkasa/Roinaa/stoat3.jpg

[[Category: Art Tutorials]]

Using the Levels Adjustment

2008-02-21T19:28:12Z

Sgt. Groovy:

The "Levels Adjustment" is divided into two modifications that it applies to an image - input and output.

===Input:===
Input reinterprets what comes in - it is used to, for example, correct a pencil drawing that should have been black lines on white paper, but scanned in as dark grey lines on light grey paper. The levels filter can stretch the luminosity (aka brightness) levels of this drawing into being black lines on white paper.

The rightmost slider controls what color, from the source image, to treat as "jet black". If your pencil lines, which were supposed to be jet black, came out as only dark grey, then you can use this to put "where black starts" just above the darkness of the lines. This makes the lines, and everything darker than they are, jet black.

The leftmost slider does the exact same thing in reverse, it's used to make everything above a certain luminosity white. This is used most commonly to chop off all the noise and texture of the paper and make it pure white.

The middle slider is a coefficient, or a "multiplier", and this means it behaves very differently than the other two. You can use the middle slider as a very crude alternative to the "curves adjustment" (and it is in fact good training for understanding what the heck the "curves adjustment" does, which isn't exactly intuitive). Simply put, the middle slider bends all the in-between shades towards either the light or dark side of the drawing - anything above it gets made light, anything below it gets made dark, and the further these are from it, the greater the effect is. Thus, if you drag it over to the far right, all of the lines will become rather dark, and if you drag it over to the far left, everything will become very light and washed out.

This can be used to adjust the contrast of your lines - it can adjust how rapidly lines fade out.

===Output:===
The output is simple - the two sliders set how dark the darkest dark is in the final picture, and how bright the brightest bright is. If you leave them both at the ends, things will range from black to white. If, for example the bottom slider were moved halfways up, nothing in the final picture would ever look brighter than 50% grey.

Other than deliberately washing out pictures, this doesn't have a lot of uses for line art.

===Example:===
In the following image, we used only the input adjustment. The white of the paper was brightened to white, and the dark was darkened to black. This isn't exactly the best demo image, since it was badly scanned in the first place, and because it was heavily JPG compressed before I got to it.

http://exong.net/wesnoth-attach/files/levels_214.png

[[Category: Art Tutorials]]

Shaped gradients with Gaussian blur

2008-02-21T14:14:06Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

The gradient tool in Inkscape is incredibly useful, but it has its limitations as well. For example, the radial gradient is limited to circular or oval in shape. To make gradients of other shapes, the Gaussian blur tool becomes handy. When you blur any object different in colour than its background, you are essentially creating a shaped gradient between those colours.

To make a "radial" gradient of any shape, that is, a gradient that changes from the edge of the path toward the centre, first create the path and set its fill colour to what you want the outer colour to be. Then make "Path -> Linked Offset", raise it to the top and set it to the inner colour (or usually somewhat more saturated one, to compensate the mixing when blurred). Use the node handle to make it smaller than the original path (1) and blur it to create the gradient (2). Because the inner path is linked to the outer one, any changes to the outer path will be updated in the shape of the inner one (though you may need to readjust the blur radius if you change the shape radically). To make multi-stop gradient, simply make several linked offsets of different colours and sizes (3). You will notice that the colour transition is smoother (rounder) than in normal gradients. If you need more linear gradient, you need to make several intermediary (in size and colour) linked offset paths.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial2.png

The same method is good for making any odd-shaped gradients. Just create a stepwise gradient with several paths and blur them to make the gradient smooth. In the image below there is a stepwise gradient created with the interpolate effect (this effect is especially handy for this, because it can interpolate the colour as well). In the lower image all but the outermost path have been blurred.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial3.png

One of the limitations of the gradient tool, when applied to strokes, is that if you want the gradient to change from the inner edge to the outer, you can only do this for oval-shaped paths. The blur, together with masking, can fix this problem as well. First create the base path, then make a linked offset path for it, raise it to the top and set its fill to none and the stroke to the outer colour (1). Create another linked offset, raise it to the top and set its fill to the inner colour, stroke to none and blur it enough to create the stroke gradient (2). Group this path with the one created in step 1. Create yet another linked offset path with no fill, and the same stroke width that the one in step 1, but with white colour (3) If you want to use a stoke style (dotted line etc.) set it for this path. Use the white stroked path as a mask for the group created earlier. Now you have a stroke with a gradient, and mostly it behaves like an ordinary stroke — edit the nodes of the base path, and the stroke changes with it. Even path operations work, as long as they produce only one path (if they produce several, the gradient stroke is only applied to the topmost one).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial4.png

Variable-width strokes

2008-02-21T14:13:40Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

In SVG the strokes are always constant in width. This is good for diagrams, but makes drawings look too formal and dull. Variable width outlines can be made with the calligraphy tool, but it requires a tablet and a relatively steady hand. The drawback of this kind of lines also is that if you decide to change the shape of the underlying colour flat, you have to redraw the outline.

Howerer, with a little trickery it's possible to create variable width outline easily for a path of any shape that behaves like the normal strokes, that is, when you change the shape of the path, the shape of the outline is updated as well.

The process is outlined in the image below:

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial1.png

# Create your path with any tool you like. Set the stroke fill to none
# From "Path -> Linked Offset" create a path whose size (and shape) is linked to the original path, ''but do not change its size''. Raise this path to the top, set its fill colour to none and the stroke colour to the desired value (here it is black with reduced opacity, so that it will always show as a darker shade of the flat colour). Set the stroke width to about the sum of thinnest and thickest lines you desire.
# Move the outline path slightly to the direction you want the thinnest parts of the outline to be (here it is moved left and up). The original path should stay completely within the outline path.
# Make a clone of the original path and use that as a clip path on the outline path. This will make invisible the parts of the outline path that lie outside the original path.

Now you have an outline that will adjust to any changes in the size or shape of the original path, and whose thickness is determined by the path normal. You can use gradient fills for the line and line styles, like dotted line etc., but remember that only the inside half of the line is visible (normal strokes lie half outside and half inside of the path).

The downside of this method is that the line thickness and shape are determined automatically and are not always exactly what you would want. To make the outline more editable, select it and first do "Path -> Object to Path" and then "Path -> Stroke to Path". The inner egde of the outline path is now a subpath that you can edit as you whish, but the line isn't linked to the original path anymore, and so doesn't keep up with any of its major changes. It's still clipped by the clone of the original path, so it will adjust to small tweakings.

Z-order tricks

2008-02-21T14:13:12Z

Sgt. Groovy:

''Note: This tutorial was written for [http://www.inkscape.org Inkscape], but should be applicable for other vector drawing programs with similar functionality.''

In a vector image, every object (path, rectangle, piece of text, etc) has a position in the x-y plane, but it also has a position in the depth dimension. If you think the objects as paper cut-outs, any time two of them overlap, one must be on top of the other. Similarily, each object in a vector image occupy a slot in the z-order, which dictates which object appears on top of which (the difference to paper cut-outs is that they retain their z-position even when they are not overlapping in x-y space). You can always rearrange the z-order of your objects, within certain limits. For example, when you group objects, they retain their relative z-order to each other, but no object outside the group can occupy a z-slot between the ingroup objects.

You can use the z-ordering to you advantage, when you want areas of different colour to appear next to each other without a gap in between. As with paper cut-outs, the simplest way to do this is just to make one area to expand a bit under the other. This has the benefit that you can later adjust the positions of the paths without a gap appearing.

This approach is not possible in all situations, however. One common situation is where you have objects with a circular z-order: Object A covering part of object B, which covers part of object C, which in turn covers part of object A. In a vector image, the z-order is always linear, and to achieve the appearance of circular z-order, we need some trickery, the tools for wich are found in the Path menu.

In the image below we have an illustration of the problem. We want three ellipses of different colours stacked circularily as in top right, but with a linear z-order, the top left is best we can do. We can't put the red ellipse under the yellow one without putting it also below the green one, because that is on top of the yellow.

http://koti.welho.com/thonkasa/Roinaa/burgshot13.png

In order to achieve the desired effect, we need to cut off a piece out of the red ellipse that corresponds to the part that would be under the yellow one. To do this, you will first select the yellow ellipse, do "Edit -> Copy" and "Edit -> Paste in place." This will result in a copy of the yellow ellipse in exactly the same place. Then select both the copy and the the red one, and do "Path -> Difference", which will remove from the red ellipse the area overlapping the yellow one. In the bottom image is a copy of of the ellipses, with the red one moved outwards to show how it's shape is changed. This also demonstrates the drawback of the method: After you have matched the edges, you can't move the paths in relation to each other without breaking the alignement.

There is another, a bit more complex way to do this operation, which will allow the paths to be edited or moved without breaking the alignment. First select the red path and make "Path -> Linked Offset", creating a copy of that will change when the original is changed. Raise it to the top and change its fill colour to white. Do the same for the yellow path, but change its colour to black. Select both linked offsets and group them. Select both the group and the red path, and do "Object -> Mask -> Set." After this operation the red path is masked with its own linked copy in white, combined a linked copy of the yellow path in black. The result is that the red path is always invisible where it overlaps the yellow one. You can edit the nodes of either path, and the mask is updated as well. You should note, however, that when a masked object is transformed (moved, scaled, etc.), the transformation is also applied to the mask, which is not desireable in our case. So if you want to do transformations on the red path, disable the mask first, do the transformation and reapply the mask (you can move it in node editing mode without this hassle, though, by simply selecting all the nodes).

[http://tavmjong.free.fr/INKSCAPE/MANUAL/html/Paths-Combining.html Here] is a summary of the path operations and how they work. They can be incredibly powerful once you learn how to use them creatively.

Let's now look at one more, a bit trickier case. Sometimes you can have a circular z-order between ''two'' objects. A good example is Saturn and it's rings, coming from behind the planet and then going over it. We see it in the image below, bottom right is our goal, and top left are the original paths, created with the ellipse tool (now you should already know how to cut the "hole" in the blue one).

http://koti.welho.com/thonkasa/Roinaa/burgshot5.png

We know that we can use a copy of the red path to cut off the part of the blue one that should be eclipsed, but how do we do that without cutting off the part that is supposed to eclipse the red path? First make "Edit -> copy" and "Edit -> Paste in place" for the red path and select this copy and the blue path. But then, instead of the "Path -> Difference" do "Path -> Division", which will cut the blue path into several paths along the red path's edge, but leave them all in place (top right). One of the parts corresponds to the section that should be covered by the red path (bottom left). Select this path and delete it. The remaining paths now make up the part that should be visible, but it is still divided into separate paths. Select them all and do "Path -> Union," which will fuse them back into one path again.

Create Art

2008-02-21T14:03:52Z

Sgt. Groovy: New vector tutorials

{| style="float:right; margin-left:1em;"
|__TOC__
|}

== Getting Started ==

=== General Information ===

Graphic artists usually meet on the [http://www.wesnoth.org/forum/viewforum.php?f=9 artwork development forum] or on the [http://www.wesnoth.org/forum/viewforum.php?f=18 restricted art development forum]. The former is a great place to post and discuss new and current Wesnoth art and graphics, and the latter to see what the art development team is working on.

Unit and terrain art is stored in the portable network graphics (PNG) format. Each frame of a unit animation, and each variation of a terrain is stored as a separate .png file in the "images" subdirectory of wesnoth, and generally these files will be 72 x 72 pixels (the size of Wesnoth's basic hexagonal tile) with an alpha channel (a part of the file that indicates how transparent each pixel is). When creating your own images, you can test them without overwriting any game data by putting them in your userdata directory (see [[EditingWesnoth]] for details on Wesnoth's directory structure). The game also supports JPEG images, though these are better suited for story art.

To edit these graphics, you'll need some program capable of creating PNGs - some of the programs in the following list are free, open-source software, and will do the job nicely: [[Art Programs]]

If you need some inspiration, we have a [[GraphicLibrary|Graphics Library]] which collects art posted on the forum. You can use this for ideas, and as a scrap heap for different parts of unit images (a technique described [[Give Your Hero A Personality|here]]).

=== Roadmap ===
A list of what's being done and what needs doing:
* '''[[Tiles Status]]''' - a roadmap/plan of sorts for future work on terrain tiles
* [http://www.wesnoth.org/forum/viewtopic.php?t=9162&sid=21cf168d2b077d06bb0017c059b4add6 A list] of current work that needs to be done with sprites.

== Art Tutorials ==

These are a work-in-progress, and describe both how to make art fit into wesnoth's style, as well as giving some considerable tips on drawing in general. Especially useful is the [[External Tutorials]] page which lists a large number of art tutorials available on the web.

=== General Art and Computer Graphics ===
* [[Using the Levels Adjustment]] - making scanned pencil drawings ''not'' look washed out.
* [[Extending dynamic range]] - The Grooviest (so far) tutorial about extending the dynamic range of images and how this technique can be used to make better scans of pencil drawings.
* [[Scanning with camera]] - How to transfer real-life art to computer using a digital camera instead of a scanner.
* [[How to Anti-Alias Sprite Art]] - A means of removing the jagged edges on pixel lines
* [[Art Supplies]] - What physical items you need to do larger cell-shaded art like that of Jetryl/Jormungadr/et al
* [[Inking With Pencils|Computer Inking a Sketch]] - Info from Jason Lutes on his portrait workflow
* [[Scaling Digital Images]] - how to properly resize an image on a computer
* [[How to Shade]] - at attempt at tackling a very complicated topic
* [[Cartography for Wesnoth|How to make Wesnoth-style Maps ]] - Kestenvarn's tricks of the trade
* [[Designing weapons and armour]] - Advice from zookeeper on designing realistic weapons for your characters
* Sgt. Groovy's vector workshop - tips and tricks for drawing with Inkscape
** [[Z-order tricks]] - few methods for faking overlapping shapes
** [[Variable-width strokes]] - how to make the strokes vary in width, like being drawn with a flat-tipped pen — no tablet needed!
** [[Shaped gradients with Gaussian blur]] - how to make gradients in other shapes than linear or radial
** [[Making portrait art in vector]] - a complete tutorial for making Wesnoth unit portraits in vector graphics.

=== [[Terrain|Terrain Graphics]] ===

The following is information specific to drawing terrain for Wesnoth. Read Frame's "Tiles Tutorial" for a good overview of how terrain graphics work.

* [[Tiles Tutorial]] - Frame's tutorial describing the process of making terrain tiles in wesnoth, and how they interact with adjacent tiles.
* [[How To Make Seamless Tiles]] - The tutorial is aimed at Photoshop users, but the technique is similar with The GIMP.
* [[Seamless Tiles Using Inkscape]] - This tutorial teaches a method for making seamless hex tiles in vector craphics (to be rendered in raster).
* [[Turning Square Tiles into Hex]] - Nifty tricks for transforming square (or any rectangle) shaped seamless tiles into hexagon seamless tiles.
* [[CastleTutorial|Castle Tutorial]] - A description of how Wesnoth's castle tiles work (needs updating, but useful nonetheless)
* [[MultiHexTutorial|Multi-Hex Tiling Tutorial]] - A description of how multi-hex tiles work.
* [[Editing Castles]] - Instructions for how to make/edit castles (and other corner-based terrains) using yobbo's GIMP script.

These describe the system used to specify how terrains behave in game:
* [[TerrainCodesWML]] - A list of the letters used to represent terrain types.
* [[TerrainGraphicsWML]] - If you really need to get technical, start here.
* [http://www.anathas.org/ayin/wesnoth/doc/terrain_graphics_wml Ayin's Terrain Graphics document] - If you really, ''really'' need to get technical, this describes the terrain graphics WML system in depth.

=== Sprite Art ===
The following are different tutorials about sprite work compiled by various wesnoth sprite artists. These will give you the most specific-to-wesnoth information about making sprites, and are well worth a read.

* [[Creating Unit Art]] - a list of specifications you will need to match.
* [[Give Your Hero A Personality]] - tricks for editing existing images, including some clip art.
* [[Basic Animation Tutorial]] - or "How to Animate Sprites for Dummies," covering the basic theory, and all of the mistakes to avoid.
* [[Advanced Animation Tutorial]] - how to wire in the animation frames to units in a way that makes the animation look good in-game. Doesn't cover any drawing as such, but focuses on timing conventions and some standard techniques for getting as much as possible out of the frames that are already drawn.
* [[Team Color Shifting]] - how to use our new team color system.
* [[Creating Shadows Under Units]] - how we create the shadows for the units in-game
* [[Making Bow Animations]] - the current standard for how we want bow animations to work
* [[Rotate Pixel Art Without Blurring]]
* [[Creating a scratch built sprite]] - an attempt to show some ways creating a sprite from scratch
* [[How to create motion blurs]]- a simple explanation on how to create attack animation weapon blurs

=== External Tutorials ===
The following page contains dozens of links to tutorials covering all manner of artwork, including sprite art. These were not made by wesnoth artists, but should prove very useful for general instruction.

* [[External Tutorials]]

== See Also ==
* [[Create]]
* [[EditingWesnoth]]
* [[GraphicLibrary]] - Lots of usercontributed graphics (most should be GPL'ed)
* [[Project]]
* [http://wesnoth.dbzer0.com/blog/wpg2 External Graphic Library] - A project to better organize the art of Wesnoth.

Shaped gradients with Gaussian blur

2008-02-21T14:02:26Z

Sgt. Groovy:

The gradient tool in Inkscape is incredibly useful, but it has its limitations as well. For example, the radial gradient is limited to circular or oval in shape. To make gradients of other shapes, the Gaussian blur tool becomes handy. When you blur any object different in colour than its background, you are essentially creating a shaped gradient between those colours.

To make a "radial" gradient of any shape, that is, a gradient that changes from the edge of the path toward the centre, first create the path and set its fill colour to what you want the outer colour to be. Then make "Path -> Linked Offset", raise it to the top and set it to the inner colour (or usually somewhat more saturated one, to compensate the mixing when blurred). Use the node handle to make it smaller than the original path (1) and blur it to create the gradient (2). Because the inner path is linked to the outer one, any changes to the outer path will be updated in the shape of the inner one (though you may need to readjust the blur radius if you change the shape radically). To make multi-stop gradient, simply make several linked offsets of different colours and sizes (3). You will notice that the colour transition is smoother (rounder) than in normal gradients. If you need more linear gradient, you need to make several intermediary (in size and colour) linked offset paths.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial2.png

The same method is good for making any odd-shaped gradients. Just create a stepwise gradient with several paths and blur them to make the gradient smooth. In the image below there is a stepwise gradient created with the interpolate effect (this effect is especially handy for this, because it can interpolate the colour as well). In the lower image all but the outermost path have been blurred.

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial3.png

One of the limitations of the gradient tool, when applied to strokes, is that if you want the gradient to change from the inner edge to the outer, you can only do this for oval-shaped paths. The blur, together with masking, can fix this problem as well. First create the base path, then make a linked offset path for it, raise it to the top and set its fill to none and the stroke to the outer colour (1). Create another linked offset, raise it to the top and set its fill to the inner colour, stroke to none and blur it enough to create the stroke gradient (2). Group this path with the one created in step 1. Create yet another linked offset path with no fill, and the same stroke width that the one in step 1, but with white colour (3) If you want to use a stoke style (dotted line etc.) set it for this path. Use the white stroked path as a mask for the group created earlier. Now you have a stroke with a gradient, and mostly it behaves like an ordinary stroke — edit the nodes of the base path, and the stroke changes with it. Even path operations work, as long as they produce only one path (if they produce several, the gradient stroke is only applied to the topmost one).

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial4.png

Variable-width strokes

2008-02-21T13:54:23Z

Sgt. Groovy:

In SVG the strokes are always constant in width. This is good for diagrams, but makes drawings look too formal and dull. Variable width outlines can be made with the calligraphy tool, but it requires a tablet and a relatively steady hand. The drawback of this kind of lines also is that if you decide to change the shape of the underlying colour flat, you have to redraw the outline.

Howerer, with a little trickery it's possible to create variable width outline easily for a path of any shape that behaves like the normal strokes, that is, when you change the shape of the path, the shape of the outline is updated as well.

The process is outlined in the image below:

http://user.personal.fi/atk/phpopintopiiri/kuvia/tutorial1.png

# Create your path with any tool you like. Set the stroke fill to none
# From "Path -> Linked Offset" create a path whose size (and shape) is linked to the original path, ''but do not change its size''. Raise this path to the top, set its fill colour to none and the stroke colour to the desired value (here it is black with reduced opacity, so that it will always show as a darker shade of the flat colour). Set the stroke width to about the sum of thinnest and thickest lines you desire.
# Move the outline path slightly to the direction you want the thinnest parts of the outline to be (here it is moved left and up). The original path should stay completely within the outline path.
# Make a clone of the original path and use that as a clip path on the outline path. This will make invisible the parts of the outline path that lie outside the original path.

Now you have an outline that will adjust to any changes in the size or shape of the original path, and whose thickness is determined by the path normal. You can use gradient fills for the line and line styles, like dotted line etc., but remember that only the inside half of the line is visible (normal strokes lie half outside and half inside of the path).

The downside of this method is that the line thickness and shape are determined automatically and are not always exactly what you would want. To make the outline more editable, select it and first do "Path -> Object to Path" and then "Path -> Stroke to Path". The inner egde of the outline path is now a subpath that you can edit as you whish, but the line isn't linked to the original path anymore, and so doesn't keep up with any of its major changes. It's still clipped by the clone of the original path, so it will adjust to small tweakings.

Z-order tricks

2008-02-21T11:27:28Z

Sgt. Groovy:

In a vector image, every object (path, rectangle, piece of text, etc) has a position in the x-y plane, but it also has a position in the depth dimension. If you think the objects as paper cut-outs, any time two of them overlap, one must be on top of the other. Similarily, each object in a vector image occupy a slot in the z-order, which dictates which object appears on top of which (the difference to paper cut-outs is that they retain their z-position even when they are not overlapping in x-y space). You can always rearrange the z-order of your objects, within certain limits. For example, when you group objects, they retain their relative z-order to each other, but no object outside the group can occupy a z-slot between the ingroup objects.

You can use the z-ordering to you advantage, when you want areas of different colour to appear next to each other without a gap in between. As with paper cut-outs, the simplest way to do this is just to make one area to expand a bit under the other. This has the benefit that you can later adjust the positions of the paths without a gap appearing.

This approach is not possible in all situations, however. One common situation is where you have objects with a circular z-order: Object A covering part of object B, which covers part of object C, which in turn covers part of object A. In a vector image, the z-order is always linear, and to achieve the appearance of circular z-order, we need some trickery, the tools for wich are found in the Path menu.

In the image below we have an illustration of the problem. We want three ellipses of different colours stacked circularily as in top right, but with a linear z-order, the top left is best we can do. We can't put the red ellipse under the yellow one without putting it also below the green one, because that is on top of the yellow.

http://koti.welho.com/thonkasa/Roinaa/burgshot13.png

In order to achieve the desired effect, we need to cut off a piece out of the red ellipse that corresponds to the part that would be under the yellow one. To do this, you will first select the yellow ellipse, do "Edit -> Copy" and "Edit -> Paste in place." This will result in a copy of the yellow ellipse in exactly the same place. Then select both the copy and the the red one, and do "Path -> Difference", which will remove from the red ellipse the area overlapping the yellow one. In the bottom image is a copy of of the ellipses, with the red one moved outwards to show how it's shape is changed. This also demonstrates the drawback of the method: After you have matched the edges, you can't move the paths in relation to each other without breaking the alignement.

There is another, a bit more complex way to do this operation, which will allow the paths to be edited or moved without breaking the alignment. First select the red path and make "Path -> Linked Offset", creating a copy of that will change when the original is changed. Raise it to the top and change its fill colour to white. Do the same for the yellow path, but change its colour to black. Select both linked offsets and group them. Select both the group and the red path, and do "Object -> Mask -> Set." After this operation the red path is masked with its own linked copy in white, combined a linked copy of the yellow path in black. The result is that the red path is always invisible where it overlaps the yellow one. You can edit the nodes of either path, and the mask is updated as well. You should note, however, that when a masked object is transformed (moved, scaled, etc.), the transformation is also applied to the mask, which is not desireable in our case. So if you want to do transformations on the red path, disable the mask first, do the transformation and reapply the mask (you can move it in node editing mode without this hassle, though, by simply selecting all the nodes).

[http://tavmjong.free.fr/INKSCAPE/MANUAL/html/Paths-Combining.html Here] is a summary of the path operations and how they work. They can be incredibly powerful once you learn how to use them creatively.

Let's now look at one more, a bit trickier case. Sometimes you can have a circular z-order between ''two'' objects. A good example is Saturn and it's rings, coming from behind the planet and then going over it. We see it in the image below, bottom right is our goal, and top left are the original paths, created with the ellipse tool (now you should already know how to cut the "hole" in the blue one).

http://koti.welho.com/thonkasa/Roinaa/burgshot5.png

We know that we can use a copy of the red path to cut off the part of the blue one that should be eclipsed, but how do we do that without cutting off the part that is supposed to eclipse the red path? First make "Edit -> copy" and "Edit -> Paste in place" for the red path and select this copy and the blue path. But then, instead of the "Path -> Difference" do "Path -> Division", which will cut the blue path into several paths along the red path's edge, but leave them all in place (top right). One of the parts corresponds to the section that should be covered by the red path (bottom left). Select this path and delete it. The remaining paths now make up the part that should be visible, but it is still divided into separate paths. Select them all and do "Path -> Union," which will fuse them back into one path again.

Making portrait art in vector

2008-02-21T09:57:35Z

Sgt. Groovy:

This tutorial

2008-02-21T09:50:45Z

Sgt. Groovy:

#REDIRECT [[Z-order tricks]]

Z-order tricks

2008-02-21T09:48:29Z

Sgt. Groovy:

In a vector image, every object (path, rectangle, piece of text, etc) has a position in the x-y plane, but it also has a position in the depth dimension. If you think the objects as paper cut-outs, any time two of them overlap, one must be on top of the other. Similarily, each object in a vector image occupy a slot in the z-order, which dictates which object appears on top of which (the difference to paper cut-outs is that they retain their z-position even when they are not overlapping in x-y space). You can always rearrange the z-order of your objects, within certain limits. For example, when you group objects, they retain their relative z-order to each other, but no object outside the group can occupy a z-slot between the ingroup objects.

You can use the z-ordering to you advantage, when you want areas of different colour to appear next to each other without a gap in between. As with paper cut-outs, the simplest way to do this is just to make one area to expand a bit under the other. This has the benefit that you can later adjust the positions of the paths without a gap appearing.

This approach is not possible in all situations, however. One common situation is where you have objects with a circular z-order: Object A covering part of object B, which covers part of object C, which in turn covers part of object A. In a vector image, the z-order is always linear, and to achieve the appearance of circular z-order, we need some trickery, the tools for wich are found in the Path menu.

In the image below we have an illustration of the problem. We want three ellipses of different colours stacked circularily as in top right, but with a linear z-order, the top left is best we can do. We can't put the red ellipse under the yellow one without putting it also below the green one, because that is on top of the yellow.

http://koti.welho.com/thonkasa/Roinaa/burgshot13.png

In order to achieve the desired effect, we need to cut off a piece out of the red ellipse that corresponds to the part that would be under the yellow one. To do this, you will first select the yellow ellipse, do "Edit -> Copy" and "Edit -> Paste in place." This will result in a copy of the yellow ellipse in exactly the same place. Then select both the copy and the the red one, and do "Path -> Difference", which will remove from the red ellipse the area overlapping the yellow one. In the bottom image is a copy of of the ellipses, with the red one moved outwards to show how it's shape is changed. This also demonstrates the drawback of the method: After you have matched the edges, you can't move the paths in relation to each other without breaking the alignement.

There is another, a bit more complex way to do this operation, which will allow the paths to be edited or moved without breaking the alignment. First select the red path and make "Path -> Linked Offset", creating a copy of that will change when the original is changed. Raise it to the top and change its fill colour to white. Do the same for the yellow path, but change its colour to black. Select both linked offsets and group them. Select both the group and the red path, and do "Object -> Mask -> Set." After this operation the red path is masked with its own linked copy in white, combined a linked copy of the yellow path in black. The result is that the red path is always invisible where it overlaps the yellow one. You can edit the nodes of either path, and the mask is updated as well. You should note, however, that when a masked object is transformed (moved, scaled, etc.), the transformation is also applied to the mask, which is not desireable in our case. So if you want to do transformations on the red path, disable the mask first, do the transformation and reapply the mask (you can move it in node editing mode without this hassle, though, by simply selecting all the nodes).

[http://tavmjong.free.fr/INKSCAPE/MANUAL/html/Paths-Combining.html Here] is a summary of the path operations and how they work. They can be incredibly powerful once you learn how to use them creatively.

Before we get to actual work, let's look at one more, a bit trickier case. Sometimes you can have a circular z-order between ''two'' objects. A good example is Saturn and it's rings, coming from behind the planet and then going over it. We see it in the image below, bottom right is our goal, and top left are the original paths, created with the ellipse tool (now you should already know how to cut the "hole" in the blue one).

http://koti.welho.com/thonkasa/Roinaa/burgshot5.png

We know that we can use a copy of the red path to cut off the part of the blue one that should be eclipsed, but how do we do that without cutting off the part that is supposed to eclipse the red path? First make "Edit -> copy" and "Edit -> Paste in place" for the red path and select this copy and the blue path. But then, instead of the "Path -> Difference" do "Path -> Division", which will cut the blue path into several paths along the red path's edge, but leave them all in place (top right). One of the parts corresponds to the section that should be covered by the red path (bottom left). Select this path and delete it. The remaining paths now make up the part that should be visible, but it is still divided into separate paths. Select them all and do "Path -> Union," which will fuse them back into one path again.

This tutorial

2008-02-21T09:46:36Z

Sgt. Groovy:

In a vector image, every object (path, rectangle, piece of text, etc) has a position in the x-y plane, but it also has a position in the depth dimension. If you think the objects as paper cut-outs, any time two of them overlap, one must be on top of the other. Similarily, each object in a vector image occupy a slot in the z-order, which dictates which object appears on top of which (the difference to paper cut-outs is that they retain their z-position even when they are not overlapping in x-y space). You can always rearrange the z-order of your objects, within certain limits. For example, when you group objects, they retain their relative z-order to each other, but no object outside the group can occupy a z-slot between the ingroup objects.

You can use the z-ordering to you advantage, when you want areas of different colour to appear next to each other without a gap in between. As with paper cut-outs, the simplest way to do this is just to make one area to expand a bit under the other. This has the benefit that you can later adjust the positions of the paths without a gap appearing.

This approach is not possible in all situations, however. One common situation is where you have objects with a circular z-order: Object A covering part of object B, which covers part of object C, which in turn covers part of object A. In a vector image, the z-order is always linear, and to achieve the appearance of circular z-order, we need some trickery, the tools for wich are found in the Path menu.

In the image below we have an illustration of the problem. We want three ellipses of different colours stacked circularily as in top right, but with a linear z-order, the top left is best we can do. We can't put the red ellipse under the yellow one without putting it also below the green one, because that is on top of the yellow.

http://koti.welho.com/thonkasa/Roinaa/burgshot13.png

In order to achieve the desired effect, we need to cut off a piece out of the red ellipse that corresponds to the part that would be under the yellow one. To do this, you will first select the yellow ellipse, do "Edit -> Copy" and "Edit -> Paste in place." This will result in a copy of the yellow ellipse in exactly the same place. Then select both the copy and the the red one, and do "Path -> Difference", which will remove from the red ellipse the area overlapping the yellow one. In the bottom image is a copy of of the ellipses, with the red one moved outwards to show how it's shape is changed. This also demonstrates the drawback of the method: After you have matched the edges, you can't move the paths in relation to each other without breaking the alignement.

There is another, a bit more complex way to do this operation, which will allow the paths to be edited or moved without breaking the alignment. First select the red path and make "Path -> Linked Offset", creating a copy of that will change when the original is changed. Raise it to the top and change its fill colour to white. Do the same for the yellow path, but change its colour to black. Select both linked offsets and group them. Select both the group and the red path, and do "Object -> Mask -> Set." After this operation the red path is masked with its own linked copy in white, combined a linked copy of the yellow path in black. The result is that the red path is always invisible where it overlaps the yellow one. You can edit the nodes of either path, and the mask is updated as well. You should note, however, that when a masked object is transformed (moved, scaled, etc.), the transformation is also applied to the mask, which is not desireable in our case. So if you want to do transformations on the red path, disable the mask first, do the transformation and reapply the mask (you can move it in node editing mode without this hassle, though, by simply selecting all the nodes).

[http://tavmjong.free.fr/INKSCAPE/MANUAL/html/Paths-Combining.html Here] is a summary of the path operations and how they work. They can be incredibly powerful once you learn how to use them creatively.

Before we get to actual work, let's look at one more, a bit trickier case. Sometimes you can have a circular z-order between ''two'' objects. A good example is Saturn and it's rings, coming from behind the planet and then going over it. We see it in the image below, bottom right is our goal, and top left are the original paths, created with the ellipse tool (now you should already know how to cut the "hole" in the blue one).

http://koti.welho.com/thonkasa/Roinaa/burgshot5.png

We know that we can use a copy of the red path to cut off the part of the blue one that should be eclipsed, but how do we do that without cutting off the part that is supposed to eclipse the red path? First make "Edit -> copy" and "Edit -> Paste in place" for the red path and select this copy and the blue path. But then, instead of the "Path -> Difference" do "Path -> Division", which will cut the blue path into several paths along the red path's edge, but leave them all in place (top right). One of the parts corresponds to the section that should be covered by the red path (bottom left). Select this path and delete it. The remaining paths now make up the part that should be visible, but it is still divided into separate paths. Select them all and do "Path -> Union," which will fuse them back into one path again.

Making portrait art in vector

2008-02-21T09:45:28Z

Sgt. Groovy:

This is a tutorial for making unit portraits for the Battle for Wesnoth in vector graphics. The Wesnoth portraits are made in what is called "the Lutesian style" after the comic artist Jason Lutes, who has donated many of the Loyalist portraits. The main reason for stylistic restrictions is to achieve aesthetic consistency.

The technical aspects of the Lutesian style can be summarised as follows:

* The colouring is made in "cell shading," that is, colours are applied in constant colour areas, known as "flats." Changes in tone (such as shadows and highlights) are also made the same way, instead of brush strokes or gradients.

* All the linework, instead of being black or having another colour of their own, is made in darker (or sometimes lighter) shade of the colour of the flat it is on top of.

* Adjacent flats don't need to be separated by outlines.

In addition to this, the final format of the portrait file should be a 205x205 pixel PNG with a fully black background.

Vector graphics programs can be easily used to produced this kind of images, because the flats can be easily crated and edited as paths filled with a solid colour. The linework, shadows and outlines are most conveniently made as semitransparent dark or light areas on top of the flats, because then they always follow the colour of the flats, if it is changed.

In this tutorial we use Inkscape as the program, but any vector drawing program worth its salt should have the same functionality. As a prequisite, you should be familiar with the basic concepts of vector drawing, how to make, edit and stroke paths, the fill rules (how to make transparent "holes" in paths), etc. Basically, reading the Inkscape manual and playing around with it for half an hour should be enough. You should also be able to draw and scan a pencil sketch four your portrait, or have other people's sketches available (remember to ask for permission).

Here's our example sketch, a young burgher:

http://koti.welho.com/thonkasa/Roinaa/burghersketch2.png

The blue lines show the edges of the shadows. It is convenient to make the shadow edges visible, because it is then easier to trace them (and scanners easily lose the lighter shades of pencilwork anyway), and by using a colour pencil, you make them separable from the actual linework. You can do the same thing with highlights, but use different colour than for shadows. The red arrow indicates the general direction of lighting.

The first thing to do is to trace out the flats and make outlines for those that are going to have them. As stated above, we are going to make the outlines as a separate, semitransparent path on top of the flat path, so that it gets its colour from it. Because lines differentiating two flats of different colours should have the colour of one or the other, but not both, it follows that the outer edge of the outline path must align exactly with the edge of the flat path. We will learn shortly how to accomplish this, but first read [[this tutorial|Z-order tricks]] about the z-order and the path tools.

Then it's time to get work on the image. First thing to do is to prepare the sketch. Since we must keep the sketch as the topmost object to see it as we draw, but also see the work beneath it, the background colour of the skecth must be changed into transparency. This can be done in GIMP, where the necessary tool can be found in "Layers -> Transparency -> Color to alpha." Once this is done, save it as PNG and import it into Inkscape. It will initially have some random X and Y values, so change them to 0, in the top toolbar. Open the document preferences and set the page size to match the selection. Rename the current layer to "sketch" and lock it, so that you don't accidentally move the sketch image, and it will be easier to select the objects under it. Create a new layer under the sketch layer and name it "flat." Now you are ready to trace the flats.

There are to ways to trace, with the freehand tool, where you move the cursor along the line and nodes are created on the fly, or the bezier tool, where you set the nodes explicitly. If you have a tablet and a very steady hand, you can manage with the freehand tool, but otherwise the bezier is better. The good thing about the bezier tool is that you end up with relatively few nodes, which keeps the file size smaller and makes later editing of the paths easier. Beginners usually make the beziers by first making an approximating polygon, and then coming back to round the corners. This is too inefficient, a faster and equally easy (after a bit of practice) way is the "tangential sweep." Follow the line and whenever there is a sharp corner, click on it without dragging. In between, click and drag tangentially to the line in few points along the curving line. You learn by practice where to place the sweep points. If there is a straight line between to corners, you don't need to put any nodes there.

The flats are best traced in back to front order, so that flats that are going to be the on the bottom are drawn first. It's usually good to base the z-order of the flats to the natural order of the real-life objects they represent, so that flats representing the foreground objects are on top of the flats representing background objects. There are times when you must break this order, but following it as a general rule helps to keep the drawing organised in your head.

In our image, we start with the sleeve at left. It's one of the "bottom objects," it's being covered by the torso, but doesn't itself cover anything. In the image below we see the sleeve traced with the bezier tool. Note that this is not the only, and not probably even the best, way to place the nodes. There actually isn't a "right" way to trace a shape, the only places where you ''must'' put a node are the sharp corners. You should try to get away with as few nodes as possible, though. You will also notice tha we didn't trace the line separating the sleeve and the torso. We will do it when we trace the flat for the torso, which comes on top of this one.

http://koti.welho.com/thonkasa/Roinaa/burgshot1.png

If we want the flat to have an outline, this is the time to do it, because it will then be directly on top of the flat it follows, and has the same relative z-placement as the flat.

The outline could be made with the calligraphy tool, but then we would be tracing the same lines twice. The flat could also be stroked, which would produce the outline automatically, but there are two problems. First, the stroke is always half outside and half inside the path, but we want the outer edge of our outline to follow the path exactly. Second, the stroke is always constant width, and this tends to result in dull and formalistic look, better suited to diagrams and such than artistic drawings. Even though strokes are not good as themselves, there is a handy tool "stroke to path," which makes a path out of a stroke so that the path follows the edges of the stroke, and we can use this to turn a stroke into a proper outline.

The "stroke to path" tool makes the original path disappear, so we will first do "Edit -> Copy" and "Edit -> Paste in place" for our flat. Then select the copy and disable fill and enable stroke. Set the stroke colour to black with 50% opacity (you might want to adjust this later, but these are good working settings at this point).

http://koti.welho.com/thonkasa/Roinaa/burgshot2.png

The width of the stroke should be about twice the intended mid-width of the outline. Then do "Path -> Stroke to path." Half of the outline still lies outside the path, and we will cut it away with a clip path, but wirst we will have to make the outline variable width. This is done easily by moving the outline path slightly, only so much that the edges of the underlying flat stay within it. Moving the outline makes the part that is inside the flat, that will remain after the intersection operation, to become thinner in some parts and thicker in other. As a general rule, the outline should be thinner on the lighted side of the object and thicker on the shadow side, so you should move the outline towards the light source.

http://koti.welho.com/thonkasa/Roinaa/burgshot3.png

When the inside part of the outline is the way you like, select the flat path and do "Edit -> Clone -> Clone". Select the clone and the outline path and do "Object -> Clip -> Set". What we did was using the clone of the flat as a clip path on the outline path, making the parts that are outside the path invisible. Because we use a clone, if we edit the nodes of the flat, the changes will automatically change the (visible) shape of the outline path correspondingly.

http://koti.welho.com/thonkasa/Roinaa/burgshot4.png

Since this method decides the outline width only by the angle of the outline, it doesn't always produce the desired outcome in all the places, but you can easily adjust the line width by moving the appropriate nodes of the ''inner edge'' of the outline path, inwards to thicken the line, outwards to thin it (the "sclupting" mode can become handy for this).

Proceed by this method for the flats in back-to-front order. Having the flats traced and outlined, my picture would look like this:

http://koti.welho.com/thonkasa/Roinaa/burgher1.png

Note that not all flats need an outline. Outlining usually adds a sense of weight and substance to objects, so thin, light and fluffy objects, such as hair, beard, moustache and the quill in my image, generally look better without.

Later it will be useful to have the silhouette of the whole figure in one path. Create a new layer titled "silhouette," select all the flat paths and do "Edit -> Copy." Switch to the "silhouette" layer and do "Edit -> Paste in place," followed by "Path -> Union." This will fuse all the paths into one, creating a silhouette. Turn the visibility of this layer off and lock it.

At this point it might be a good time to handle some special features, such as eyes. We could make them as flats on top of the face flat, but the pupils usually require adjustements to make them look just right, it's more convenient to make the eyes as holes in the face, and put the pupils behind it. If you make the pupils as whole ovals, only part of them is visible, but when you move them, the part always shows which should.

There are two ways to make a hole in a path. One is to make a path the shape of the hole on top of it and doing "Path -> Difference," but if you select a path and press shift while starting to make a new one, the new one will become a subpath to the selected one, and if it's inside it, it will appear as a hole (provided the fill rule for the path is even-odd, the default).

If you want the holes to have outlines as well, it will require a bit of trickery. Select the face flat and do "Edit -> Copy" and "Edit -> Paste in place." Then do "Path -> Break apart," which will make each of the subpaths into separate paths. Select the outermost path and delete it. What you are left with are paths that excactly cover the holes. Follow the outline making procedure, except in the end, instead of "Path -> Intersection," do "Path -> Difference" to clip away the ''inner'' half of the outline path.

For the eyes, you need white for the eyeballs (the white you see is the background colour). So make a eyeball colour flat (not usually pure white) that covers both of the eye holes, and drop it below the face flat. Then use the ellipse tool to make the iris and the pupil, group them, and drop the group below the face flat, but above the eyeball flat. When you have made one iris/pupil combo, you can copy it to get the other be exactly alike.

http://koti.welho.com/thonkasa/Roinaa/burgshot11.png

Now would be a good time to make the rest of the linework, that is the lines not outlining any flats. These lines are also to be made dark with partial transparency, so that they become a darker shade of the colour of the flat they are covering. There are several ways to make the linework. With short lines extending from the outlines, like the ones on the sleeves in my picture, can be made by making new nodes in appropriate places on the outline path and dragging them out. Thick lines can me treated as the flats by tracing the edges of the lines with the bezier tool. If you have a tablet, the calligraphy tool might be useful, but it takes some practice to use well. If the calligraphy lines overlap, you have to fuse them together by doing "Path -> Union" on them, or the partial transparency will produce darker shades on the overlapping parts.

Sometimes tracing linework is done more easily on a raster program, because you can use the eraser tool to trim the lines and correct mistakes immediately (especially with a tablet, when you can use the other end of the stylus as the eraser). The prosedure is as follows: Open the original sketch bitmap on a raster program. Create a new transparent layer on top and do the linework on this with black. Save only this layer as a PNG and import it into Inkscape. Set the X and Y properties of the bitmap to zeroes to make it align with the rest of the image. Select the bitmap and do "Path -> Trace bitmap" (the default options work well most of the time). Delete the bitmap and what you are left with is the linework done in paths. They are fully black so decrease the opacity and edit them further with the usual tools, if necessary.

Whatever method you use to make the linework, it's good to keep things organised and put it on its own layer.

http://koti.welho.com/thonkasa/Roinaa/burgher2.png

After the linework we can do the shadows. Create a new layer titled "shadow" on top of the others, and start tracing the shadow edges like you did with flats, but obviously, the shadows don't need outlines. You can make the shadow paths extend beyond the outlines of the figure, and cut them off in the end using the silhouette path you created earlier, using "Path -> Intersection." The shadows will also be partially transparent to show through the colour of the flat, but their own colour can well be something else than black. In my image I have used dark blue for the shadows, as in outdoors the light coming to the shadow parts comes from the sky and is bluer than the sunlight. You can try different colours for your shadows to see how it affects the general mood if the image.

http://koti.welho.com/thonkasa/Roinaa/burgher3.png

We still have some details to do, like the buttons on the jacket. Here I made the buttons with the ellipse too, using the same ellipse to create the shadow (by now you should be able to figure out how to make the crescent shape). If you need identical objects, you can always make copies of one, but if you make clones instead, and decide to change something about the design of the object later, you only need to change the original and the clones change as well.

http://koti.welho.com/thonkasa/Roinaa/burgshot12.png

To complete the image, some final touches are needed, like the specular spots in the eyes and some texture and highlights for the beard and hair (all done with semitransparent dark and light paths). Then the standard black background and some cropping and we're all done (I'll leave it as the exercise of the reader to figure out how to do the cropping).

http://koti.welho.com/thonkasa/Roinaa/burgherfinal1.png

Create Art

2007-10-02T20:15:22Z

Sgt. Groovy: /* Terrain Graphics */

{| style="float:right; margin-left:1em;"
|__TOC__
|}

== Getting Started ==

=== General Information ===

Graphic artists usually meet on the [http://www.wesnoth.org/forum/viewforum.php?f=9 artwork development forum] or on the [http://www.wesnoth.org/forum/viewforum.php?f=18 restricted art development forum]. The former is a great place to post and discuss new and current Wesnoth art and graphics, and the latter to see what the art development team is working on.

Unit and terrain art is stored in the portable network graphics (PNG) format. Each frame of a unit animation, and each variation of a terrain is stored as a separate .png file in the "images" subdirectory of wesnoth, and generally these files will be 72 x 72 pixels (the size of Wesnoth's basic hexagonal tile) with an alpha channel (a part of the file that indicates how transparent each pixel is). When creating your own images, you can test them without overwriting any game data by putting them in your userdata directory (see [[EditingWesnoth]] for details on Wesnoth's directory structure). The game also supports JPEG images, though these are better suited for story art.

To edit these graphics, you'll need some program capable of creating PNGs - some of the programs in the following list are free, open-source software, and will do the job nicely: [[Art Programs]]

If you need some inspiration, we have a [[GraphicLibrary|Graphics Library]] which collects art posted on the forum. You can use this for ideas, and as a scrap heap for different parts of unit images (a technique described [[Give Your Hero A Personality|here]]).

=== Roadmap ===
A list of what's being done and what needs doing:
* '''[[Tiles Status]]''' - a roadmap/plan of sorts for future work on terrain tiles
* [http://www.wesnoth.org/forum/viewtopic.php?t=9162&sid=21cf168d2b077d06bb0017c059b4add6 A list] of current work that needs to be done with sprites.

== Art Tutorials ==

These are a work-in-progress, and describe both how to make art fit into wesnoth's style, as well as giving some considerable tips on drawing in general. Especially useful is the [[External Tutorials]] page which lists a large number of art tutorials available on the web.

=== General Art and Computer Graphics ===
* [[Using the Levels Adjustment]] - making scanned pencil drawings ''not'' look washed out.
* [[Extending dynamic range]] - The Grooviest (so far) tutorial about extending the dynamic range of images and how this technique can be used to make better scans of pencil drawings.
* [[Scanning with camera]] - How to transfer real-life art to computer using a digital camera instead of a scanner.
* [[How to Anti-Alias Sprite Art]] - A means of removing the jagged edges on pixel lines
* [[Art Supplies]] - What physical items you need to do larger cell-shaded art like that of Jetryl/Jormungadr/et al
* [[Inking With Pencils|Computer Inking a Sketch]] - Info from Jason Lutes on his portrait workflow
* [[Scaling Digital Images]] - how to properly resize an image on a computer
* [[How to Shade]] - at attempt at tackling a very complicated topic
* [[Cartography for Wesnoth|How to make Wesnoth-style Maps ]] - Kestenvarn's tricks of the trade
* [[Designing weapons and armour]] - Advice from zookeeper on designing realistic weapons for your characters
* [[Making portrait art in vector]] - Sgt. Groovy's complete tutorial for making Wesnoth unit portraits in vector graphics.

=== [[Terrain|Terrain Graphics]] ===

The following is information specific to drawing terrain for Wesnoth. Read Frame's "Tiles Tutorial" for a good overview of how terrain graphics work.

* [[Tiles Tutorial]] - Frame's tutorial describing the process of making terrain tiles in wesnoth, and how they interact with adjacent tiles.
* [[How To Make Seamless Tiles]] - The tutorial is aimed at Photoshop users, but the technique is similar with The GIMP.
* [[Seamless Tiles Using Inkscape]] - This tutorial teaches a method for making seamless hex tiles in vector craphics (to be rendered in raster).
* [[Turning Square Tiles into Hex]] - Nifty tricks for transforming square (or any rectangle) shaped seamless tiles into hexagon seamless tiles.
* [[CastleTutorial|Castle Tutorial]] - A description of how Wesnoth's castle tiles work (needs updating, but useful nonetheless)
* [[MultiHexTutorial|Multi-Hex Tiling Tutorial]] - A description of how multi-hex tiles work.
* [[Editing Castles]] - Instructions for how to make/edit castles (and other corner-based terrains) using yobbo's GIMP script.

These describe the system used to specify how terrains behave in game:
* [[TerrainLettersWML]] - A list of the letters used to represent terrain types.
* [[TerrainGraphicsWML]] - If you really need to get technical, start here.
* [http://www.anathas.org/ayin/wesnoth/doc/terrain_graphics_wml Ayin's Terrain Graphics document] - If you really, ''really'' need to get technical, this describes the terrain graphics WML system in depth.

=== Sprite Art ===
The following are different tutorials about sprite work compiled by various wesnoth sprite artists. These will give you the most specific-to-wesnoth information about making sprites, and are well worth a read.

* [[Creating Unit Art]] - a list of specifications you will need to match.
* [[Give Your Hero A Personality]] - tricks for editing existing images, including some clip art.
* [[Basic Animation Tutorial]] - or "How to Animate Sprites for Dummies," covering the basic theory, and all of the mistakes to avoid.
* [[Advanced Animation Tutorial]] - how to wire in the animation frames to units in a way that makes the animation look good in-game. Doesn't cover any drawing as such, but focuses on timing conventions and some standard techniques for getting as much as possible out of the frames that are already drawn.
* [[Team Color Shifting]] - how to use our new team color system.
* [[Creating Shadows Under Units]] - how we create the shadows for the units in-game
* [[Making Bow Animations]] - the current standard for how we want bow animations to work
* [[Rotate Pixel Art Without Blurring]]

=== External Tutorials ===
The following page contains dozens of links to tutorials covering all manner of artwork, including sprite art. These were not made by wesnoth artists, but should prove very useful for general instruction.

* [[External Tutorials]]

== See Also ==
* [[Create]]
* [[EditingWesnoth]]
* [[GraphicLibrary]] - Lots of usercontributed graphics (most should be GPL'ed)
* [[Project]]

Turning Square Tiles into Hex

2007-10-02T20:11:46Z

Sgt. Groovy:

Turning Square Tiles into Hex

2007-10-02T19:19:18Z

Sgt. Groovy: /* Method 2: Rotation */

In Wesnoth the Terrain graphics are made with seamlessly tiling hexagonal pixmaps. These are more difficult to make than square tiles, for which ready made scripts exist in most image manipulation programs, and which can also be found in multitude of libraries, free to use.

If you cut a square tile into a hexagon, it's only going to be seamless in north and south edges. To make it seamless in all edges takes a bit of trickery, but it's relatively simple. There are two methods to do it, the one that is better depending on the circumstances.

==Method 1: Keeping the Orientation==

Let's start with a dry mud texture from the standard GIMP library:

http://koti.welho.com/thonkasa/Roinaa/hextile1.png

First it must be scaled into 54x36 pixel size (this applies to the standard 72x72 pixel Wesnoth hexes, for other hexes the figures are width*0.75 and height/2). Then resize the canvas by 2 in both directions and fill the empty space with copies of the tile:

http://koti.welho.com/thonkasa/Roinaa/hextile2.png

Now crop the image into 72x72 size (the exact placement of the cropping does not matter) and trim the corners so that it matches the standard hex:

http://koti.welho.com/thonkasa/Roinaa/hextile3.png

Now comes the moment of truth: what happens when we tile this hex? Let's see:

http://koti.welho.com/thonkasa/Roinaa/hextile4.png

Almost makes you believe in magic, doesn't it?

There are two important things to notice:

# The size and aspect ratio of the original texture are not preserved (unless it is already 54x36 pixels), but this can be a good thing. If you start with a square tile where the pattern is seen as from above, then it gets scaled down vertically, creating an impression of perspective (in Wesnoth, the terrain is seen from an angle, not directly above).
# The hex will contain the original pattern twice, that is, every unique feature in the original tile will appear in the hex tile two times. This may make the pattern to appear too densely repeating, but once you have the hex made, you can change something about one copy of each feature to counteract this effect (as long as you don't touch the edge pixels, the tile will stay seamless).

==Method 2: Rotation==

In the second method the original tile must be square, for it will be rotated 45 degrees. The rotation may prove useful in cases where the pattern have straight forms, like stripes or checkers. In that case the end result will generally look better if the lines go from SE to NW and SW to NE rather than from south to north and west to east. So if your pattern has these lines vertically and horizontally, you should use this method.

If the original tile is not square, scale it horizontally or vertically so that it is. Double the size of the canvas in both directions and fill the space with copies of the tile, creating a 2x2 pattern. Below is our example, a ceramic tile pattern:

http://koti.welho.com/thonkasa/Roinaa/hextile21.png

Next rotate the image 45 degrees and scale it so that the height becomes 144 and width 216 pixels:

http://koti.welho.com/thonkasa/Roinaa/hextile22.png

Then cut the hex shape out (anywhere where it doesn't include transparent pixels), and that's it (notice that the pattern appears only once in the hex):

http://koti.welho.com/thonkasa/Roinaa/hextile23.png

If you doubt that it can be this simple, lets see how it tiles:

http://koti.welho.com/thonkasa/Roinaa/hextile24.png

Which method to use depends on your requirements. If you need to preserve the original orientation, use method 1. If you need the rotation, or must have the pattern repeated only once per hex, use method 2 (both preserving orientation and having 1 repeat/hex is not possible with these methods). If the orientation and repeat number don't matter, the appropriate method depends on the desired size of the pattern: method 2 reproduces the pattern twice as big as method 1.

Turning Square Tiles into Hex

2007-10-02T19:12:17Z

Sgt. Groovy: Another method added

In Wesnoth the Terrain graphics are made with seamlessly tiling hexagonal pixmaps. These are more difficult to make than square tiles, for which ready made scripts exist in most image manipulation programs, and which can also be found in multitude of libraries, free to use.

If you cut a square tile into a hexagon, it's only going to be seamless in north and south edges. To make it seamless in all edges takes a bit of trickery, but it's relatively simple. There are two methods to do it, the one that is better depending on the circumstances.

==Method 1: Keeping the Orientation==

Let's start with a dry mud texture from the standard GIMP library:

http://koti.welho.com/thonkasa/Roinaa/hextile1.png

First it must be scaled into 54x36 pixel size (this applies to the standard 72x72 pixel Wesnoth hexes, for other hexes the figures are width*0.75 and height/2). Then resize the canvas by 2 in both directions and fill the empty space with copies of the tile:

http://koti.welho.com/thonkasa/Roinaa/hextile2.png

Now crop the image into 72x72 size (the exact placement of the cropping does not matter) and trim the corners so that it matches the standard hex:

http://koti.welho.com/thonkasa/Roinaa/hextile3.png

Now comes the moment of truth: what happens when we tile this hex? Let's see:

http://koti.welho.com/thonkasa/Roinaa/hextile4.png

Almost makes you believe in magic, doesn't it?

There are two important things to notice:

# The size and aspect ratio of the original texture are not preserved (unless it is already 54x36 pixels), but this can be a good thing. If you start with a square tile where the pattern is seen as from above, then it gets scaled down vertically, creating an impression of perspective (in Wesnoth, the terrain is seen from an angle, not directly above).
# The hex will contain the original pattern twice, that is, every unique feature in the original tile will appear in the hex tile two times. This may make the pattern to appear too densely repeating, but once you have the hex made, you can change something about one copy of each feature to counteract this effect (as long as you don't touch the edge pixels, the tile will stay seamless).

==Method 2: Rotation==

In the second method the original tile must be square, for it will be rotated 45 degrees. The rotation may prove useful in cases where the pattern have straight forms, like stripes or checkers. In that case the end result will generally look better if the lines go from SE to NW and SW to NE rather than from south to north and west to east. So if your pattern has these lines vertically and horizontally, you should use this method.

If the original tile is not square, scale it horizontally or vertically so that it is. Double the size of the canvas in both directions and fill the space with copies of the tile, creating a 2x2 pattern. Below is our example, a ceramic tile pattern:

http://koti.welho.com/thonkasa/Roinaa/hextile21.png

Next rotate the image 45 degrees and scale it so that the height becomes 144 and width 216 pixels:

http://koti.welho.com/thonkasa/Roinaa/hextile22.png

Then cut the hex shape out in the middle, and that's it (notice that the pattern appears only once in the hex):

http://koti.welho.com/thonkasa/Roinaa/hextile23.png

If you doubt that it can be this simple, lets see how it tiles:

http://koti.welho.com/thonkasa/Roinaa/hextile24.png

Which method to use depends on your requirements. If you need to preserve the original orientation, use method 1. If you need the rotation, or must have the pattern repeated only once per hex, use method 2 (both preserving orientation and having 1 repeat/hex is not possible with these methods). If the orientation and repeat number don't matter, the appropriate method depends on the desired size of the pattern: method 2 reproduces the pattern twice as big as method 1.

Turning Square Tiles into Hex

2007-09-30T14:53:31Z

Sgt. Groovy:

Create Art

2007-09-30T13:54:21Z

Sgt. Groovy: Adding new tutorial

{| style="float:right; margin-left:1em;"
|__TOC__
|}

== Getting Started ==

=== General Information ===

Graphic artists usually meet on the [http://www.wesnoth.org/forum/viewforum.php?f=9 artwork development forum] or on the [http://www.wesnoth.org/forum/viewforum.php?f=18 restricted art development forum]. The former is a great place to post and discuss new and current Wesnoth art and graphics, and the latter to see what the art development team is working on.

Unit and terrain art is stored in the portable network graphics (PNG) format. Each frame of a unit animation, and each variation of a terrain is stored as a separate .png file in the "images" subdirectory of wesnoth, and generally these files will be 72 x 72 pixels (the size of Wesnoth's basic hexagonal tile) with an alpha channel (a part of the file that indicates how transparent each pixel is). When creating your own images, you can test them without overwriting any game data by putting them in your userdata directory (see [[EditingWesnoth]] for details on Wesnoth's directory structure). The game also supports JPEG images, though these are better suited for story art.

To edit these graphics, you'll need some program capable of creating PNGs - some of the programs in the following list are free, open-source software, and will do the job nicely: [[Art Programs]]

If you need some inspiration, we have a [[GraphicLibrary|Graphics Library]] which collects art posted on the forum. You can use this for ideas, and as a scrap heap for different parts of unit images (a technique described [[Give Your Hero A Personality|here]]).

=== Roadmap ===
A list of what's being done and what needs doing:
* '''[[Tiles Status]]''' - a roadmap/plan of sorts for future work on terrain tiles
* [http://www.wesnoth.org/forum/viewtopic.php?t=9162&sid=21cf168d2b077d06bb0017c059b4add6 A list] of current work that needs to be done with sprites.

== Art Tutorials ==

These are a work-in-progress, and describe both how to make art fit into wesnoth's style, as well as giving some considerable tips on drawing in general. Especially useful is the [[External Tutorials]] page which lists a large number of art tutorials available on the web.

=== General Art and Computer Graphics ===
* [[Using the Levels Adjustment]] - making scanned pencil drawings ''not'' look washed out.
* [[Extending dynamic range]] - The Grooviest (so far) tutorial about extending the dynamic range of images and how this technique can be used to make better scans of pencil drawings.
* [[Scanning with camera]] - How to transfer real-life art to computer using a digital camera instead of a scanner.
* [[How to Anti-Alias Sprite Art]] - A means of removing the jagged edges on pixel lines
* [[Art Supplies]] - What physical items you need to do larger cell-shaded art like that of Jetryl/Jormungadr/et al
* [[Inking With Pencils|Computer Inking a Sketch]] - Info from Jason Lutes on his portrait workflow
* [[Scaling Digital Images]] - how to properly resize an image on a computer
* [[How to Shade]] - at attempt at tackling a very complicated topic
* [[Cartography for Wesnoth|How to make Wesnoth-style Maps ]] - Kestenvarn's tricks of the trade
* [[Designing weapons and armour]] - Advice from zookeeper on designing realistic weapons for your characters
* [[Making portrait art in vector]] - Sgt. Groovy's complete tutorial for making Wesnoth unit portraits in vector graphics.

=== [[Terrain|Terrain Graphics]] ===

The following is information specific to drawing terrain for Wesnoth. Read Frame's "Tiles Tutorial" for a good overview of how terrain graphics work.

* [[Tiles Tutorial]] - Frame's tutorial describing the process of making terrain tiles in wesnoth, and how they interact with adjacent tiles.
* [[How To Make Seamless Tiles]] - The tutorial is aimed at Photoshop users, but the technique is similar with The GIMP.
* [[Seamless Tiles Using Inkscape]] - This tutorial teaches a method for making seamless hex tiles in vector craphics (to be rendered in raster).
* [[Turning Square Tiles into Hex]] - A nifty trick for transforming square (or any rectangle) shaped seamless tiles into hexagon seamless tiles.
* [[CastleTutorial|Castle Tutorial]] - A description of how Wesnoth's castle tiles work (needs updating, but useful nonetheless)
* [[MultiHexTutorial|Multi-Hex Tiling Tutorial]] - A description of how multi-hex tiles work.
* [[Editing Castles]] - Instructions for how to make/edit castles (and other corner-based terrains) using yobbo's GIMP script.

These describe the system used to specify how terrains behave in game:
* [[TerrainLettersWML]] - A list of the letters used to represent terrain types.
* [[TerrainGraphicsWML]] - If you really need to get technical, start here.
* [http://www.anathas.org/ayin/wesnoth/doc/terrain_graphics_wml Ayin's Terrain Graphics document] - If you really, ''really'' need to get technical, this describes the terrain graphics WML system in depth.

=== Sprite Art ===
The following are different tutorials about sprite work compiled by various wesnoth sprite artists. These will give you the most specific-to-wesnoth information about making sprites, and are well worth a read.

* [[Creating Unit Art]] - a list of specifications you will need to match.
* [[Give Your Hero A Personality]] - tricks for editing existing images, including some clip art.
* [[Basic Animation Tutorial]] - or "How to Animate Sprites for Dummies," covering the basic theory, and all of the mistakes to avoid.
* [[Advanced Animation Tutorial]] - how to wire in the animation frames to units in a way that makes the animation look good in-game. Doesn't cover any drawing as such, but focuses on timing conventions and some standard techniques for getting as much as possible out of the frames that are already drawn.
* [[Team Color Shifting]] - how to use our new team color system.
* [[Creating Shadows Under Units]] - how we create the shadows for the units in-game
* [[Making Bow Animations]] - the current standard for how we want bow animations to work
* [[Rotate Pixel Art Without Blurring]]

=== External Tutorials ===
The following page contains dozens of links to tutorials covering all manner of artwork, including sprite art. These were not made by wesnoth artists, but should prove very useful for general instruction.

* [[External Tutorials]]

== See Also ==
* [[Create]]
* [[EditingWesnoth]]
* [[GraphicLibrary]] - Lots of usercontributed graphics (most should be GPL'ed)
* [[Project]]

Turning Square Tiles into Hex

2007-09-30T13:53:15Z

Sgt. Groovy: New tutorial

Create Art

2007-09-20T17:08:42Z

Sgt. Groovy: /* Terrain Graphics */ Adding new tutorial

{| style="float:right; margin-left:1em;"
|__TOC__
|}

== Getting Started ==

=== General Information ===

Graphic artists usually meet on the [http://www.wesnoth.org/forum/viewforum.php?f=9 artwork development forum] or on the [http://www.wesnoth.org/forum/viewforum.php?f=18 restricted art development forum]. The former is a great place to post and discuss new and current Wesnoth art and graphics, and the latter to see what the art development team is working on.

Unit and terrain art is stored in the portable network graphics (PNG) format. Each frame of a unit animation, and each variation of a terrain is stored as a separate .png file in the "images" subdirectory of wesnoth, and generally these files will be 72 x 72 pixels (the size of Wesnoth's basic hexagonal tile) with an alpha channel (a part of the file that indicates how transparent each pixel is). When creating your own images, you can test them without overwriting any game data by putting them in your userdata directory (see [[EditingWesnoth]] for details on Wesnoth's directory structure). The game also supports JPEG images, though these are better suited for story art.

To edit these graphics, you'll need some program capable of creating PNGs - some of the programs in the following list are free, open-source software, and will do the job nicely: [[Art Programs]]

If you need some inspiration, we have a [[GraphicLibrary|Graphics Library]] which collects art posted on the forum. You can use this for ideas, and as a scrap heap for different parts of unit images (a technique described [[Give Your Hero A Personality|here]]).

=== Roadmap ===
A list of what's being done and what needs doing:
* '''[[Tiles Status]]''' - a roadmap/plan of sorts for future work on terrain tiles
* [http://www.wesnoth.org/forum/viewtopic.php?t=9162&sid=21cf168d2b077d06bb0017c059b4add6 A list] of current work that needs to be done with sprites.

== Art Tutorials ==

These are a work-in-progress, and describe both how to make art fit into wesnoth's style, as well as giving some considerable tips on drawing in general. Especially useful is the [[External Tutorials]] page which lists a large number of art tutorials available on the web.

=== General Art and Computer Graphics ===
* [[Using the Levels Adjustment]] - making scanned pencil drawings ''not'' look washed out.
* [[Extending dynamic range]] - The Grooviest (so far) tutorial about extending the dynamic range of images and how this technique can be used to make better scans of pencil drawings.
* [[Scanning with camera]] - How to transfer real-life art to computer using a digital camera instead of a scanner.
* [[How to Anti-Alias Sprite Art]] - A means of removing the jagged edges on pixel lines
* [[Art Supplies]] - What physical items you need to do larger cell-shaded art like that of Jetryl/Jormungadr/et al
* [[Inking With Pencils|Computer Inking a Sketch]] - Info from Jason Lutes on his portrait workflow
* [[Scaling Digital Images]] - how to properly resize an image on a computer
* [[How to Shade]] - at attempt at tackling a very complicated topic
* [[Cartography for Wesnoth|How to make Wesnoth-style Maps ]] - Kestenvarn's tricks of the trade
* [[Designing weapons and armour]] - Advice from zookeeper on designing realistic weapons for your characters
* [[Making portrait art in vector]] - Sgt. Groovy's complete tutorial for making Wesnoth unit portraits in vector graphics.

=== [[Terrain|Terrain Graphics]] ===

The following is information specific to drawing terrain for Wesnoth. Read Frame's "Tiles Tutorial" for a good overview of how terrain graphics work.

* [[Tiles Tutorial]] - Frame's tutorial describing the process of making terrain tiles in wesnoth, and how they interact with adjacent tiles.
* [[How To Make Seamless Tiles]] - The tutorial is aimed at Photoshop users, but the technique is similar with The GIMP.
* [[Seamless Tiles Using Inkscape]] - This tutorials teaches a method for making seamless hex tiles in vector craphics (to be rendered in raster).
* [[CastleTutorial|Castle Tutorial]] - A description of how Wesnoth's castle tiles work (needs updating, but useful nonetheless)
* [[MultiHexTutorial|Multi-Hex Tiling Tutorial]] - A description of how multi-hex tiles work.
* [[Editing Castles]] - Instructions for how to make/edit castles (and other corner-based terrains) using yobbo's GIMP script.

These describe the system used to specify how terrains behave in game:
* [[TerrainLettersWML]] - A list of the letters used to represent terrain types.
* [[TerrainGraphicsWML]] - If you really need to get technical, start here.
* [http://www.anathas.org/ayin/wesnoth/doc/terrain_graphics_wml Ayin's Terrain Graphics document] - If you really, ''really'' need to get technical, this describes the terrain graphics WML system in depth.

=== Sprite Art ===
The following are different tutorials about sprite work compiled by various wesnoth sprite artists. These will give you the most specific-to-wesnoth information about making sprites, and are well worth a read.

* [[Creating Unit Art]] - a list of specifications you will need to match.
* [[Give Your Hero A Personality]] - tricks for editing existing images, including some clip art.
* [[Basic Animation Tutorial]] - or "How to Animate Sprites for Dummies," covering the basic theory, and all of the mistakes to avoid.
* [[Advanced Animation Tutorial]] - how to wire in the animation frames to units in a way that makes the animation look good in-game. Doesn't cover any drawing as such, but focuses on timing conventions and some standard techniques for getting as much as possible out of the frames that are already drawn.
* [[Team Color Shifting]] - how to use our new team color system.
* [[Creating Shadows Under Units]] - how we create the shadows for the units in-game
* [[Making Bow Animations]] - the current standard for how we want bow animations to work
* [[Rotate Pixel Art Without Blurring]]

=== External Tutorials ===
The following page contains dozens of links to tutorials covering all manner of artwork, including sprite art. These were not made by wesnoth artists, but should prove very useful for general instruction.

* [[External Tutorials]]

== See Also ==
* [[Create]]
* [[EditingWesnoth]]
* [[GraphicLibrary]] - Lots of usercontributed graphics (most should be GPL'ed)
* [[Project]]

Seamless Tiles Using Inkscape

2007-09-20T17:07:48Z

Sgt. Groovy: new tutorial

Designing weapons and armour

2007-08-01T15:32:27Z

Sgt. Groovy:

by Zookeeper

Often one sees that someone draws these things looking all silly. Slight exaggeration of weaponry and such is of course fine in a fantasy setting, but it's not uncommon at all to go over the board and draw something that couldn't function in reality at all. This mostly applies to portraits and other realistic imagery, not really unit graphics as such.

Now, my point is simply: you can make weapons and armour be flashy and cool while keeping them very realistic, and a real-world reference also helps you to get things functional without having to think what kind of sword handle would work and what wouldn't, for example (and, no offense to artists drawing fantasy stuff, but that usually goes a bit awry), and that there are plenty of references available. So even if you don't feel the need for the weapons you draw to look realistic, having some references still helps to actually make it look good (in the same way as having a reference when drawing a person makes it much easier to get it look right and good).

For this purpose, I thought I'd go through some related sites I know of and post some links here. These are armouries, general information/review sites, etc, which contain good pictures of historical arms and armour. Mostly (if not all) are only european stuff though (which sort of fits with at least the mainline Wesnoth setting), but that's all I had available right now.

* http://www.armouronline.com/ - the full body armour section is good, weapons are generally crappy here.
* http://www.myarmoury.com/ - hard to navigate, but contains a lot of good pictures, if you take the time to find them. The "reviews" page is a good place to start from.
* http://www.volny.cz/grex/vyrobky/vyrob_e.htm - some odd czech site, but has some nice armour.
* http://www.lutel.cz/index_.php?en=1 - quite a number of all sorts of weapons in the catalog (especially the polearms and such look nice).
* http://www.darksword-armory.com/ - a collection of often historically accurate weaponry
* http://www.by-the-sword.com/ - various things
* http://www.siue.edu/COSTUMES/history.html - a site with lot of pictures of clothing from different eras and areas, from 19 century engravings. The scans could be of higher quality, but you can find plenty of stuff for reference/inspiration for fantasy clothing. Many of the same pictures can also be found in a book ''A Pictorial History of Costume'' (Pepin Press, no author).

Feel free to post your additions (especially to the type of items not covered on these sites), too, since we don't have much here yet.