Trims panels in UDIMs best way to Substance Painter?

Artistic Support Materials and Textures
1

image
image487×905 34.6 KB

I have a few trim sheet panels that I want to bake and detail in Substance Painter;

  • all the texture sets will have to be in a single material
  • I want to have them sorted vertically in in the UV space
  • I know how to bake UDIMs on a single object already, but I’m a bit confused when it comes to trim sheets and UDIMs.

Questions:

  • is it best to bake each panel individually and then import each texture set in the final UDIM in Blender, or do I export from Blender all the highpoly in a single FBX, and the lowpoly in a single FBX, UV unwrapped to the UDIM tiles, and texture it all together?
  • if I go with the first way, are there any blunders in the importation that I have to be aware of?
  • if I go with the second way, is best that the lowpoly is merged in a single mesh in Blender, before exporting, or I can export them in a single FBX but as separate meshes (is there a difference?)

Thanks!

2

On your third question, I measured it. Three panels filling tiles 1001, 1002 and 1003, corners parked on exactly u = 1.0 and 2.0, out to FBX and back in.

Separate objects and one joined mesh both came back with max UV drift 0.000000000 and no boundary corner off its integer. Same on 3.6.23, 4.2.23, 4.5.12, 4.5.13, 5.2.0 and 5.2.1. The joined file was smaller, 12188 bytes against 17212.

So joining is a workflow choice, not a data one. If a tile lands wrong later, the round trip isn’t where it happened. I planted a 1e-6 nudge first, to check the comparison could see one.

3

I ended going with a joined workflow, for practicality in managing the files, but there is a major hassle in that, with the ID map ending having same colors on different parts.

In Blender, I assigned a “random” face / vertex color per any object composing the panel to be trimmed, but the process evidently doesn’t distribute well the color values, or Painter doesn’t interpret them well when creating the ID map from them, which result in having to use at least 3 different masking method, in that app, per area.

So, probably, the separated method is better, even also on a iteration-safe point of view.