Hey guys
I've just created my first human body model and i need to unwrap it but i'm not sure what methods or uv layout is best for use in games. I have attempted some research but i still don't have an example from in the industry. Any help on this would be great.
Replies
pelt mapping is great if you don't care about wasted uv space, but i'm fairly confident if i just pelt mapped and sent the result over to a client, i wouldn't be asked to work for them again.
The idea is to break the entire UV shell into pieces that will:
A.) Attempt to get every polygon face in the UV layout as close to 1:1 with it's
corresponding mesh polygonal face.
B.) Fit everything in the 1 quadrant of the UV space you're entitled to
C.) Not put seams in areas where the texture should look continuous
D.) Give enough space/resolution to the UV shell bits that will draw the most
attention to themselves (usually the face is in that category.)
* You can have multiple UV sets, but sometimes attempting to set up multiple UV sets for a mesh in one app and then sending the mesh, plus UV sets to another package for rendering, animation etc. can create problem since all apps can read data in your file other than the basic mesh in different ways.
I've never understood why someone that does training doesn't do a DVD on unwrapping several different body types with different costumes and other things like multi-part meshes, curled tentacles, bent multi-opposite joint articulating arms, legs etc. I would have suffered so much less UV pain if something like that had been around.
Another thing...just so you know, everyone does it differently but many times you can pelt the face and maybe the feet, use cylindrical projection for the arms and legs (or pelt the legs) and then I usually use an orthographic view for the top and bottom of the hands and then sew each hand together on the pinky side. A lot of times I can get away with pelt mapping the front and back of what's left of the torso.
To add to this if you keep getting significant distortion in some models it can be helpful to UV the whole thing via Ptex, projection paint everything in 3D then project the non-distorted results to your layed out UVs for any additional PS work you might have to do.
I also don't understand why no software has integrated some automatic stacking options in with the UV editing tools either...sometimes perfectly symmetrical portions of the model can be stacked right on top of each other after you flip one side to match. I guess it's fine to get visual cues for overlapping maps even when your intending to overlap, but it would be nice to not have that and to also have mathematically correct UV shell positioning for stacked UV production vs. having to eyeball it.
Not only is it important to maintain a 1:1 ratio within your unwrap but try to maintain a consistent texel density between various objects and the game world.
So for example 512px per meter or some similar formula that works for that game/engine, like 4px per every game unit or whatever works. That way you don't have ultra sharp heads on 2048px sheets sitting in a world with 512px covering an entire wall and a soda can using a 2048 sitting on a 128px floor tile. Try to get these things within a reasonable range. Pixel grids are great for this, a lot of engines have texel modes so that can really help.