3D Questions and Answers

If you’re thinking of going with a single map like I usually do for low draw calls in small mobile games, the main things to think about are seams, distortion and spacing. 

A good UV map should be easy enough to interpret without the model which means logical seams, have enough space between shells for pixel bleed (having some unused space in the map is almost inevitable)and has minimal distortion.

Example: Character UV (below). The unwrap is easy enough to read, uses up the most space it can. Has a bigger uv for the face for the most detail, and in this case doesn’t have overlapping UVs - although here’s a link about it being totally acceptable to do. 

Nowadays we don’t need to worry too much about blatantly visible UV seams thanks to programs like substance painter and blender allowing us to paint diffuse maps on the models themselves. However we do still need to be smart about the unwrapping as a whole. 

Large shells usually mean less seams and less seams mean it’s easier to read what the map shows. Finding where the character’s legs are compared to their arms to add a graze or a freckle can be a 5 second job or a 5 minute job with multiple programs open depending on the layout of the character’s UV map.  If you’re on a tight schedule an easy to read map is a must.

“ UV Shells are UV islands. When you unwrap a model, you usually 'break it up' in smaller pieces (unwrap modifier). [These] unwrapped pieces are UV shells (or islands) “

- Zbuffer

Hidden seams are definitely going to be a benefit when planning out UV shells. Normal maps can sometimes distort along the seams when baking in some programs but that’s a whole other kettle of fish. Just know that UV seams are going to be a big part of baking, if you’re doing that. And most people like to put seams on hard edges and where clothing seams usually are.

See here for a good forum post on UV seams and Normal Maps. 

Distortions in the UV map that are extreme will likely result in smeared pixels. Obviously this will affect how sharp the texture map looks on the model itself. “Relaxing” the UVs usually takes care of this in your chosen 3d program.

UV mapping grids like the one below can really help visualize the distortions that are being made by the UVs. 

Spacing and pixel bleed has been covered really well here it’s definitely something to keep in mind. I suggest a little plug in from Flaming Pear called solidify to help with this. It clamps the textures in Photoshop. I highly recommend deleting any colours you have in the free space around the UV shells. 

Paul H. Paulino has made an excellent article on their website for texture mapping tips, you can visit his page here

So, in some cases UVs can be spread over several maps. See this video. The guys from Flipped Normals explain that to paint high detail (say for VFX), you’ll need a ridiculously high pixel texture map so the solution to not killing your cpu while opening the project is to separate the UVs into UDIMs. (I highly recommend their videos)

“ UDIM stands for U-Dimension. UDIM is a way of creating a single linear number that identifies each integer block in UV space. You do this by defining a limit on the number of patches you can use in the U direction (the U-DIM) and then resetting U to 0 and incrementing V by one after reaching that number “

- fxguide

This is also pretty useful for different materials: e.g wood goes in map 1, glass goes into map 2, stone goes into map 3, etc. I personally don’t have a need to do this for every project but it’s something that a lot of people do.

Example image from Allegorithmic’s help page

As you can see in the above image, the spacing between the UV shells are optimized with the ‘gun’ and ‘legs’ UDIMs leaving very little space between shells.

Also see Texture Atlas/Decal sheet/packed texture