Hello,
Does anyone here use Allegorithmic substance Designer or Bitmap2Material in their pipeline? I’m interested to hear what you have to say about it - i.e. are you actively using it, is it good, does it save you time?
Thanks!
I was just looking at the cg-talk thing last night: http://www.cgsociety.org/index.php/CGSFeatures/CGSFeatureSpecial/allegorithmic
It seemed to have come a long way as a texturing tool, though the videos seem to only show objects using 0-1 UVs.
I haven’t found also any metrics regarding the processing cost for generating their textures at runtime (granted it’s a possible problem mostly for current-gen). Or how does that work for streaming worlds (as opposed to generating textures only at load time) etc…
Either way, it seems worth looking into the trial.
I’ve always liked the product – it’s the sort of thing that appeals to the TA mindset – but historically they’ve had a basic strategic problem getting adoption.
They never really solved runtime procedural expansion on 360/PS3 class hardware – procedural textures could be expanded pretty quickly but they had to go into conventional video memory after they were generated, so there was no win in terms of limited vram. It was cool that it worked at all - but effectively this meant that the whole thing amounted to a very efficient disk space compression method – nice, but not something that most developers really needed (it is, however, why many of their early adopters were big PC rpg games where runtime memory was not an issue but download time file sizes were.
As a content creation tool, I like it – i’ve been a fan of procedural texturing forever (going back to DarkTree, PowerAnimator and RenderMan). If you have people on staff who’ll run with it they can be incredibly productive in the ‘generically similar but different in detail’ kind of texturing. The new version includes the ability to drive procedures with mesh data (for example, pick a color based on the curvature of a surface) which addresses what used to be their big weakness relative to Maya node networks.*
So, overall it’s a handy thing to have around, but not the procedural panacea we’d like to see. I doubt that will change in the DX 11 generation, although maybe the combination of more direct download games and more (though never enough!) runtime memory might make it a more attractive economic proposition.
**Disclaimer – that’s hearsay, no personal use with the current version
Theodox - that sounds really interesting. But it seems to confirm my suspicions - that its true power lies in combination with an engine. I was asked if it would speed up our outsourcing workflow where the textures requested by clients are usually TGA’s, PSDs, DDSs or something like that. So even if we use Substance, we would still have to “bake out” the procedural textures to deliver them, which I think defeats the purpose. For our co-dev projects it may be a nice tool though.
One other thing that bothers me a bit is that the community seems to be rather small - the Allegorithmic forums aren’t exactly overflowing with knowledge. Even PSN Home’s forums (and PSN Home is rather specialized!) are more active. So I’m also a bit worried it’s not one of the products where people buy licenses because it looks useful but then never actually use it.
If you’re a max-primary shop, it’s still probably a decent win in the sense that you can create a large variety of visually distinct textures fast (dunno about formats but I’d assume that’s a solved problem). If you’ve already go sophisticated maya users, 85% of what they do for offline texture generation can be done in Maya with shading networks (though they do have some neat noise primitives that Maya doesn’t have, and there are some image processing functions that don’t have obvious maya analogs).
i have only used substance to create tiling texture maps.