Bones can be much lighter, unless you do something ridiculous like give each vert it's own bone.
The kinds of skinning techniques shown in the video have been an active area of academic research for over a decade now. There is a lot of published work out there for anyone who would like to develop similar tools.
Err, I don't get it, he is modeler,texturer AND a professional rigging artist?
I never thought it was possible to master all these at this level. He must've a lot behind him.
The hard part of the technology behind these tools is done and published by people with advanced degrees in computer science and related fields. As long as you have the patience and dedication to invest in reading and understanding the research papers you can make tools like this.
Also the researchers I have ever contacted about their work have been more than happy to assist with implementation, even if the paper goes back years. They won't do the work for you, but they will answer questions.
Anthony is too kind as always! (and a brilliant artist to boot)
Purely execution-wise joints might be a bit more expensive because it's another layer between the vert and the transformation. But! A lot of performance these days boils down to how fast you can shuffle data around. And with next gen it's a lot more data. Having lets say 900 blend shapes for a 50 000 vert face mesh means 45 million positions to be stored and that's not even mentioning the normals. So if you can animate the same face with a 100 bones that's unimaginably less data.
Err, I don't get it, he is modeler,texturer AND a professional rigging artist?
I never thought it was possible to master all these at this level. He must've a lot behind him.
Ya he's a fucking wizard. Definitely an exception to the rule. I'm guessing he put the time in to learn all of that and works efficiently.
And yea, like the posted said above, knowing how to code or having a CS degree seems to help a lot. There are a few modellers that have their own scripts they whip up to solve some problem they have in their pipeline. Just being able to do that alone is pretty awesome, let alone the stuff that Andrew does.
Replies
[ame]https://www.youtube.com/watch?v=fDtL3PpNI-g[/ame]
[ame]https://www.youtube.com/watch?v=HQZw7XxCLmM[/ame]
A lone artist DID do those tech demos, the grass one and such. His name is Andrew Maximov (d1ver) amazingly talented guy
However wouldn't it be a massive resource hog to have all of those bones instead of a vertex animation?
The kinds of skinning techniques shown in the video have been an active area of academic research for over a decade now. There is a lot of published work out there for anyone who would like to develop similar tools.
I never thought it was possible to master all these at this level. He must've a lot behind him.
The hard part of the technology behind these tools is done and published by people with advanced degrees in computer science and related fields. As long as you have the patience and dedication to invest in reading and understanding the research papers you can make tools like this.
Also the researchers I have ever contacted about their work have been more than happy to assist with implementation, even if the paper goes back years. They won't do the work for you, but they will answer questions.
Purely execution-wise joints might be a bit more expensive because it's another layer between the vert and the transformation. But! A lot of performance these days boils down to how fast you can shuffle data around. And with next gen it's a lot more data. Having lets say 900 blend shapes for a 50 000 vert face mesh means 45 million positions to be stored and that's not even mentioning the normals. So if you can animate the same face with a 100 bones that's unimaginably less data.
thanks, haiddasalami! I'd love to if I can
Ya he's a fucking wizard. Definitely an exception to the rule. I'm guessing he put the time in to learn all of that and works efficiently.
And yea, like the posted said above, knowing how to code or having a CS degree seems to help a lot. There are a few modellers that have their own scripts they whip up to solve some problem they have in their pipeline. Just being able to do that alone is pretty awesome, let alone the stuff that Andrew does.
http://livestream.com/gnomon/naughty-dog
Thanks to Andrew (and any others on the panels if you're reading this!) Really inspiring stuff!