I've been trying to read on it online, but everything's so technical or vague or both that I can't wrap my head around the concept/s: what, in layman's terms are diffuse maps, specular maps, bump maps, etc? And what are the differences? how are they different from texture maps?
I'm a programmer so I'll tackle this from a technical point of view yet still try to be clear...
Let's begin with diffuse maps. These are your normal textures that 3d game engines have used ever since they've been around. Each pixel in the texture (often referred to as a texture pixel or texel) lets the engine know what the diffuse color of the object the texture is mapped to is. Imagine a model that is just a wall, a diffuse map would probably contain red bricks etc. If the game engine then draws the model, it will select that texture and apply that to the wall and you'll get your textured wall. Games soon wanted better lighting effects so up to games such as Half-Life and the same generation, artists would "bake" lighting details, shadows, reflections etc. into the diffuse texture, meaning they would probably add shadows to that red-brick-wall-texture to simulate a light etc. This works but with programmable graphics pipeline (via shaders), newer games now want to calculate all the lighting and shading dynamically so that, if for example a light moves, the shadows in the texture map move as well. If you bake those details into a diffuse map, it wouldn't work, the lighting would have to be static or look fake.
The other types of maps are usually telling the engine things it needs to know in order to calculate details on the fly. There are different names and methods around but let me try to cover the grounds here:
Normal maps (often called bump maps too) are these blue-purpleish textures you might have come across before (google image search otherwise). They contain information as to what the normal at a specific spot looks like. If you don't know what a normal is, educate yourself somewhere. Anyways, the normal vectors x, y and z components are stored in the image as the r, g and b channel. As you know, the r, g and b values can range from 0-255, so a normal vector pointing straight up for a flat surface would be 0,0,1 in vector components and 127,127,255 in rgb values. That's why these are so purple, because most normals there point up more or less. The engine then knows for each texel (definition above) what the normal would look like and thus it can fake shadows and lighting.
Specular maps just let the engine know how shiny a certain part of the texture is. Often these are grayscale and a brightness of 0 means that this texel isn't shiny at all, 255 means it's as shiny as can be. So you could have a wood door with a metal handle in one diffuse map and then draw a specular map where all the parts of the wood are almost black and the metal parts are white, like a mask, and, in-game, the wood wouldn't reflect specular light, the door handle would.
Hope this makes it clearer, ask again if you still have questions, I have some additional links and images I could prepare.
