:sigh:
Here is a quick-n-sleazy example to get my point across. After this, I'm done.
1. Mapping channel 1. Just your regular unwrap job on a cylinder.
2. Mapping channel 2. Took the quads in the middle of the cylinder and made them bigger than what they are in channel 1. This is so the bump map in that area can have more pixels devoted to them than in channel 1. In a way, this is like making that particular part resolution independant. Filled with a simple dot pattern, but I'm not showing it.
To start, a regular white texture on the model. Add lights to the scene and maybe use some photometrics. For this, just two basic lights with nothing fancy. Bake channel 1 and save it.
Then add the bump map and make sure it is in channel 2. Bake it out to channel 1 and save it.
Open them up in Photoshop and get busy. Slap 'em together, use masks, paint some colour in, play with blending modes, and all that jazz. Since this is a simple example, nothing too fancy. Save it and apply.
Two ideas are going on here:
1. Bake for lighting. This goes for the model in general and for the bump map. That is, once the bump map is baked in, there is no longer a need for a bump map. Plus it's an easy way to keep the lighting consistant across the entire model. This is particularly true for curved surfaces (like cylinders).
2. Channels to make different parts resolution independant. I didn't up size the bump very much, but I could have gone much bigger for even finer detail (if the final texture size will allow for it).
Now, resolution independant is a poor phrase for what I'm talking about. This is more of a reference to the final texture dimensions and uv chunks. If you think about it in that context, it might make sense.
I'm telling you, this is the way to go. It might seem like more work, but it's not in the long run. You'll be saving yourself a lot of headache.
Time to digress.
Later, ya'll.