I have been trying to work with these texture all day to get something that looks better then what I had. I was reading this tutorial...
http://www.citiesxl.com/index.php?/content/view/36/2/lang,en/
and when I saw the color map, I noticed there was not a lot of detail included. Some variations, a little dirt, but for the most part looks almost like a vector image. The normal bump map gives a lot of the actual definition on the geometry. Well I took this idea and kind of ran with it, I took all of the textures, and vectored them out using the cutout filter in PS CS3. Here are three renders I took.
Color Bump and Diffuse maps.
Bump and Diffuse maps.
Color and Diffuse map.
Old Color maps-also in this scene the bump setting is pretty low.
I can post the actual different maps if you would like for me to, I just thought this would be ok to do. I guess the question I really am wanting to ask, should my color maps look like above(vectored like on top) and let the diffuse and normal map create the light/dark variations, or should the have more detail in them(Dirt and grime like the bottom)? In the screen with all three maps, I like somethings more, like the air ducts. I like the columns on the right, not the left, but I think that is because there is only one light source in these renders.
Replies
If you're aiming for a more realistic style, use photographs as blending layers or points of reference for painting in detail by hand.
Edit: Look at the thread directly below this one, "10 texturing techniques", the CGSociety article - you can probably pick up some useful techniques there.
The diffuse contains the colour information of the surface and usually there's at least a wee bit of implied surface texture in there, too, to help things along. But it's very rare that a surface is completely uniform blocks of colour (doubly so in games because that usually looks kinda boring). There's often discolourations, dirt, marks, grime, grain, dust, etc. no matter how faint.
You can see in the example you linked to that although there's large areas of fairly flat colour, there's still a lot of fine detail and variation, which (partly because they're taken from photos) gives it that realistic look.
When you posterise the diffuse textures like you've done, you're sacrificing surface detail for a sort of cel-shaded look. What was an intricately varied bit of rusting with flecks of orange and brown and red on your vents (i.e. lots of detailed colour information) becomes a flat orange blob. If that's the artistic style you're after, then that's cool. But if you're after something realistic, it's not how things look in real life.
It might be best to think of everything else as being there to enhance/support the diffuse maps, rather than be a replacement for detail in them. It's not very often that you put detail into those maps that isn't already displayed in some way on your diffuse texture.
In general, if a model looks good with no lighting or textures on it other than the diffuse colour (fullbright) it'll tend to look good with lighting and shader textures (assuming, of course, that they're made well).
1. What different types of information should be on the different maps of mine. I would create a texture map, apply that to the object. Then I would take that image black and white it out to create a "diffuse" map and then take that black and white image and create a normal bump map out of that. Then attach all three images to the geometry. All three images have the same image. They are just in different colors.
EG
This is the normal bump map used.
http://www.citiesxl.com/images/stories/blog/General/winner08.jpg
This is the color map used.
http://www.citiesxl.com/images/stories/blog/General/winner11.jpg
Where as I would take the second image and create the normal and diffuse maps from that. So the better question would be, what information should be on which maps? Like...the grout and grates in the bump normal map, and the rust and water marks on the color map?
Also if the diffuse and color map are the same, why are there different places to put different files for the different things?
I read that article and that is what made me look for more tuts around here, and then I found the thread tutorial sticky list or whatever and looked at the last one. Which is the one I posted.
There is still SOOOOO much I have to learn its intimidating!
Taking the diffuse texture and making a normal map straight out of that isn't usually good enough. It creates incorrect normal maps.
I wrote a mini-tutorial about making normal maps when you only have a diffuse texture that might be of some use.
In general...
Diffuse map contains surface colour and brightness. Maybe some ambient occlusion shadowing and a tiny bit of lighting information. If you're not using normal maps, it will also contain all of the lighting information (shadows and highlights).
Normal map contains normal information baked from a high-res source (i.e. you can make a 6000 poly model look like it's 2 million poly counterpart) and/or it also contains normal detail that's been generated from a bump map (stuff like rough surface on stone, wood grain, pores on skin, etc... things that it's a pain to model yet).
Specular controls the brightness and colour of the specular highlight on the surface.
Gloss (or exponent or specular power) controls how diffused (i.e. how sharp/intense or broad/dull) the specular highlight is.
Glow (or luminence or illumination) controls how much and what colour the surface glows.
There are other maps, but they're probably the most common you'll come across.
I would suggest that you may benefit from practicing with getting diffuse-only textures nailed first and then try your hand at ones with normal and spec, etc... You might find you get a better idea of what the extra textures can do for a model once you understand what bits they replace in the diffuse texture and how they do it better.