Multi tile uv's

EnV is right, you can paint in Internal with GLSL no problem across the multiple images (don’t forget to save each as you finish painting) but I think the problem with painting in Cycles is that you still have to select the image node to make it active, I have to try with that again to check, though I usually do all my painting in Internal material texture stack so I can use GLSL for feedback.

@cymps Hey, I tried doing this but I can only create tiles in 1 direction X or Y. Can you please re-post this image in higher resolution or add blend file so I can see what are your mapping values? Thanks!

I´m using Blender 2.79b daily dev, but I can´t seem to MIX the maps between 1001 and 1002.
The entire map for the character I´m using has the head on 1001 and the body on 1002.
So how do I mix this on a single shader? Using mix, will provide a factor: ghosting 50% of the face and 50% of the body. Not good.
Am I supposed to have multiple shaders? one for 1001, then another one for 1002?
Any help would be appreciated.
Thanks @pitiwazou itiwazou

Use the group nodes I added on the post :wink:

Yeah, I see it working on the diffuse part, what about doing the same for speculars, sss and normal maps?

I wish we could bake to UDIM tiles and have a native support for UDIMs :frowning: If you want movie guys to use Blender you want UDIM support, because movie stuff means huge textures with UDIM tiles, also Katana people like it. I am regularly asked for this and I end up going back to Modo or Mudbox so I can do proper UDIM texture painting and exports and bakes etc

2 Likes

I’ve heard talk about UDIMS being implemented sometime after 2.8

Not sure, but maybe, hopefully?

Hope so. You can work around multi tiles but the issue main issue is being able to bake an object’s textures to multiple UDIM tiles, as far as I can tell that requires quite a bit work work at the moment, so I personally avoid doing it in Blender.

Using characteristics already existing in Blender, is there an addon that allows import our file, tick a box “Import uDims as UV Maps”, and it will import all of your tiles and rename them according to their UDIMs ?

Then also when you’re done and exporting, tick the box “Export UV Maps as uDims”, this will keep the UDIM imported tiles and the names assigned to them upon import.

I dont care if It stacks all UV maps into one UV space. It just need to keep tiling data of coordinate space at import/export.

Thx

I am seeing high resolution texture maps of human bodies. SO is the purpose of splitting this up in tiles only to save on CPU use ? for Games. ? Lke it just seams really weird that we are taking all the parts and turning them and trying to make images for them. It seams like total kayos. I do not understand why a person would not keep everything on one tile and make that one tile bigger and keep all the unwraps together as if they were physically unwrapped. Why have the belly turned upside down next to the chest or have the eyeball way far away from the cheek. Why not keep the eyeball and cheek together on one image ?

It’s not about saving on CPU use at all. Originally this workflow was created for VFX, where you would want very high resolution maps. Most render engines won’t allow for crazy image resolutions like, say, a single 32k image, while they would allow multiple 4k or 8k images. Another consideration is level of detail, where you can have lower detail (smaller UVs) eyeball on one tile and higher detail eyball for close-up shots on its own, dedicated, tile. Plus, you get somewhat easier modularity - using different images for particular pieces without having to substitute the texture for the entire model.
Lastly, the most trivial example: say you just aren’t getting enough detail with a single 2k image. But you would with two of them. Your options then are:

  • use a 4k square image, quadrupling your memory usage, although you really only need half of it
  • use a single non-square 4k by 2k image
  • use two square 2k images

Memory-wise, second and third options are the same. But the simplest to deal with is the third, since when dealing with non-square textures you’ll have squished UVs and would have to take into account this 2:1 aspect ratio issue.
This can of course be done without UDIMs, using multiple UV maps and multiple materials, and in fact this is what OPs setup does. UDIMs allow this to be more efficient (single material, single texture lookup, etc.)