I've just watched over the Substance Days talks that were held at Gnomon last year.
One of the character artists mentioned using UDIM's which I think is just breaking a model up into multiple UV squares and assigning different material ID's?
Is this good practice for portfolio work?
I'm currently creating the LP+UV's for my Panzer IV tank and was wondering if I should do the same or whether to just bake it all down to one 4096x4096 texture.
Could anyone explain what UDIM's actually are and when there use is applicable?
Thanks!
Replies
https://www.fxguide.com/featured/udim-uv-mapping/
The advantage over putting everything in just one tile is that you can handle much more data this way. Rather than working with one 32768x32768 texture you can use 256 2048x2048 textures.
For portfolio work you want to have enough texels that you'll have about one texel per pixel in all of the shots you're rendering. Figure out what the closest close-up shot will be and figure the texel density should be from that. If you're just rendering at 1024x600 for WAYWO you don't really need that many texels and you can straightforwardly use just one UV tile. If you're rendering a close-up of The Illusive Man's eyes for an opening scene in Mass Effect you might need to use a few tiles to get the amount of texel density you need to get a good result.
For actual games you generally standardize on a certain amount of texel density per unit, since players can potentially go anywhere and lick any wall. You can get more texel density out of certain textures by stacking shells, using tilable textures, and so on, making the game run and look better on systems that couldn't run the game otherwise. That doesn't matter as much for portfolio work, since that only has to render on one machine and it can take as long to render as you want it to.
That said it can be a good idea to show that you can get the most out of the UV space you use by stacking/straightening/packing as necessary.
The point of UDIM is to unwrap outside the 0-1 space for more density, correct ?
You can do this in games already by using multiple material ids.
Then you can have as much texel density on each UV set as you want.
Why is this workflow desired or is just the buzzword that gets everyone hot and bothered.
UDIM seem totally irrelevant for games.
i.e: skin material with its own 4K texture space and material/metal/leather/chipped paint/whatever/etc.....
Udims where 'sort of' added to substance Painter but are not fully implemented and lack features that you use Udims for in the first place.
Here is a vid I put up recently on my YT channel showing Udims in Max 2017 to S Painter. I don't use this workflow but thought it might be helpful for people like yourself unfamiliar with Udims. They've been around for years in VFX workflow and were specifically invented to support the massive amounts of texture data that a single asset needs in VFX. Something that will never be applicable to games, as the current state of the tech stands.
https://www.youtube.com/watch?v=1BplWdNkmbs
So there's no performance gain using UDIM, compared to just assigning 1 material 1 uvmap like the traditional way?
Btw I know this is an old thread. Can substance paint on multiple texture sets at once in the latest version?
but i use it also for chars... 3-5 udims to paint and bake all in one go...
for feature film its more or less impossible to work without it... you cant assign 200 textures everytime you change something...
or open up 200 files in photoshop only to adjust the saturation...
same for the viewport in mari at the time. zoomed out just a little it looked like some fuzzy game screenshot with a far to aggressive mipmap bias setting applied. not really how i'd like to work.
btw. mudbox naming convention for texture tiles is better IMO. no idea why they standardized on this instead.
i have not touched compositing in a good while but i think in shake you just told the program where to look for the sequence by feeding it special characters somehwere within the filename, like # or @.