single or multiple objects into one uv map?

uodna25

I'm a student learning 3d modeling for games in Korea, particularly for environment assets. I've learned that it's good to combine multiple objects and unwrap them into one uv map whenever possible, for optimization and to control consistent texel density, rather then creating uv maps per object. For example, like combining a drum can, tire, and a glass bottle into one uv map (looks like they call this a uv atlas, not sure).

Now im going into substance painter, and it's starting to get confusing. Say that I sculpted a gun in zbrush. All i learned is to retopo, unwrap, then texture, bake the maps in substance painter, creating maps just for that gun. There just seems no room for the 'combining objects into one uv' process to fit in the pipeline. I've looked through a bunch of youtube videos, but they all create a map for just that single object.

My question is, when to combine the uvs of objects (before or after using substance painter), or do I don't need to be overly possessed with it? How do AAA games go over with this. I'm really confused because I never seen a single tutorial in youtube texturing multiple objects in substance painter. 

Thank you in advance!

Eric Chadwick

Hey!

This is indeed a useful skill, and we do have lots of resources to help explain it (and many experienced artists here who do it!).

An atlas is one way to do this kind of workflow, where all the UVs are contained within the 0,1 uv space. 

You can mirror and stack UVs to reuse space, for example to UV both left and right arms to the same texture space.

This is often called Modular workflow. 

http://wiki.polycount.com/wiki/Modular_environments

Another related strategy is called Trim Sheets. Where your UVs can be tiled across one axis, and thus go outside the 0,1 space.

Usually this UV work is done in dedicated 3d modeling software, like Max or Maya or Blender. Because Painter really isn’t setup for UV editing.

I recently made a quick video of how UVs can be reused easily, see https://polycount.com/discussion/238306/how-do-people-get-so-much-resolution-in-their-textures-on-low-poly-models/p1

Eric Chadwick

uodna25 said:

For example, like combining a drum can, tire, and a glass bottle into one uv map

It’s good you mentioned this specific example. 

Objects can be setup to share UV space specifically so they can reuse the same material. This optimizes the rendering speed, by reducing draw calls.

However, anything using transparency is usually separated from things without, and uses a dedicated material. 

This is because transparency is difficult in a real-time renderer, and often requires some careful consideration. 

Rendering performance, sorting problems, and if you use the alpha channel that impacts texture memory.

So, the glass bottle would usually use a different UV layout and material than the can and tire.

gnoop

It was a very common technique once  with combining  multiple neighboring objects in a single, often non-square texture  for a sake of  having one shared distant LOD object in systems using hierarchical lods  or lod grouping .    Still works with that logic  and  good for mobile games for sure  but AAA  games moved toward UDIM  materials or combination of repeating  and unique parts  where sometimes a single set of textures  is not enough for a single object anymore.    

Also in S Painter you can do textures  for  some master UDIM   unwrap optimized for quick hires  texturing and then   rebake the result for ingame  UVs  with multiple objects packed , uv space priority , non-square formats, whatever your project needs.