Hi, I'm not an artist, although I'm pretty creative for a programmer. I want to help my friends (who are the real artists) with making 2D game contents. As you might know, there's often a gap between the technical and the art side, while making a game. My friends know how to draw/photoshop/styles, etc. But they don't really know how to make images suitable for a game engine. I think I should be the bridge between the tech and art department, so I was wondering if you guys have tips. We are making a game (hobby project for now). The world contains a lot of creepy hospital/industrial indoor scènes. One of my concerns is how to get all those textures? We need all kind of surfaces, wet floors, tiled walls, concrete walls, dirty walls, lightsources, bricks, and so on. The point is that they need good quality and additional maps as well. The engine I made is capable of doing all kinds of stuff (normalMapping, HDR, reflections, radiosity, and so on), so it would be a shame if we didn't use that. But that means we need high quality textures, specular maps, normalMaps, etc. I know how to make them (technically), so I can explain them details. But in the end they need to produce a library with all those textures. Where to start? Howto make them? By hand/photography? There are free sources on the web, but ussually they aren't that good, or they miss additional textures. And maybe... if the project gets commercial some day, we need our own textures anyway I guess. How do the artists of, let's say Halflife2, create all those textures? Greetings, Rick

There are a variety of ways to make textures, including all those maps you need (diffuse, specular, normal, opacity, parralax).

From photos- undoubtedly the easiest, but also the most limiting. Photographic textures have a distinct photographic feel, usually, but for environments they are usually fine. However, finding/taking good photos is a challenge. 3dtotal.com has excellent texture CDs (I recommend volumes 1 and 2 highly), and Marlin Studios has good texture CD's as well. Best is to take photos yourself, though, on an overcast day without much sunlight.
However, when you need to extract normal and specular maps from them, its a real pain. You're going to have to paint a grayscale bump map by hand, and convert that to a normal map. You are also going to need to paint your specular map by hand, instead of just changing the saturation and levels of individual diffuse layers.

By hand (1)- Many games, especially during the time of Half-Life and Doom3, seemed to paint all their textures literally pixel-by-pixel. No photoshopped photos, no texture generators, just hard work and sweat. This is extremely hard to do, but results in a very high quality texture. However, once you have things like normal and specular maps, this sort of diffuse detail is a thing of the past, and will actually hurt the final product.

By hand (2)- This is how I make most of my textures. I'll use photos or texture generators (Alienskin EyeCandy), about 10 layers and different blending modes, to create a 'base' texture. I then paint on top of that, either with custom brushes or dodging and burning.

A good texture is good because conceptually, the artist is strong. The reason photo textures are popular is because they are 'natural' already. You don't need to know where to paint rust heavier on an iron item, its already there. You don't need to study how oxidation grows on copper, its already there. You don't need to study how wood gets rotted, its already there. And ultimately this familiarity is what makes photo textures less desirable than hand-made textures.
Furthermore, specular maps are extremely important to giving your surface a 'realistic' look. These take practice, as there are few in-depth sources about them, but given their subtle importance, there should be more. For normal maps, www.bencloward.com is a must-read.
Ultimately, your artists will need to learn the technical side. Which is extremely easy compared to the art side.

Thanks professor!

I really thought most games would using photograps, but it turns out different. You made an important note about the combination of normalMaps/specularMaps and diffuseMaps. When taking a photo, it most probably contains shadows and highlights because the surface is litten by something. But those shadows/hightlights shouldn't be there of course, since the normalMap/specularMap is going to do that. Thanks for telling that!

Greetings,
Rick