Walt Disney's ElasticSkin

I don’t know if you’ve seen Walt Disney’s ElasticSkin presented on Siggraph 2011:

[quote=““Siggraph 2011 Paper””]
We present a new algorithm for near-interactive simulation of skeleton driven, high resolution elasticity models. Our methodology is used for soft tissue deformation in character animation. The algorithm targets performance through parallelism using a fully vectorized and branch-free SVD algorithm as well as a stable one-point quadrature scheme on a hexahedral grid.
[/quote]

Here’s a video-demonstration: http://www.disneyanimation.com/library/elasticSkinning-movie.mov
The PDF paper: http://www.disneyanimation.com/library/elasticSkinning.pdf
And some tech notes: http://www.disneyanimation.com/library/elasticSkinning-supplemental.zip

This looks so friggin awesome! Do you think this will be released to the public, or is this something that will be kept inhouse?

This is really awesome. The people who made this are part of Joseph Teran’s Research Group. http://www.youtube.com/user/TeranGroup/videos

I am not sure of the extent to which they are directly associated with Disney though it mentions they worked with them for this research.

It is a set of techniques and solvers, so I doubt it will be “released” but the ideas will hopefully trickle down to us soon! :slight_smile:

Thanks for the links. I hadn’t seen the paper!

Cool Stuff! Thanks for the share :D:

Really good stuff! Can’t wait to hear more from these guys.

Wish this was ready when we were doing Tangled, would’ve saved us months on shot cleanup …

This looks absolutely fantastic in terms of results. I would however would like to know what is the workflow they are envisioning for it. Similar finite element solutions we’ve explored would require a complete departure from traditional rig set up (instead you set up underlying drivers for the cage and etc) Would be interesting to see if they have developed a way to use this as a post deformer.

Thnxs 4 sharing! Great Stuff here!

just to round this thread out… here the link to the complete disney animation library of publications and papers:

http://www.disneyanimation.com/library/list.html

crazy stuff, i am telling you… :slight_smile:

Walt Disney’s ElasticSkin is really cool

and thanx shcmack and also mattschwarz for sharing links

and just wan to ask do any have information about Disney’s Student Development program because i am wishes to apply for Character Artist after i finished my Rigging classes.

[QUOTE=mattschwarz;13491]just to round this thread out… here the link to the complete disney animation library of publications and papers:

http://www.disneyanimation.com/library/list.html

crazy stuff, i am telling you… :)[/QUOTE]

Browsing through the paper and watching the video, I’d say that their own Y position in the video on “Compute Expense” is totally incorrect.

AFAIK Pose-based deformers (which it shares Y position with) can be run at realtime in many situations, while theirs clock in at 5.14s per frame on an NVIDIA Quadro 6000…

I’d also be wary as there seems, like with dual quaternion skinning, to create other artifacts that you’d have to deal with. First there is the possibly negative values that seem to be produced with their stretch, and then there are some square shape on the fingers which isn’t really explained. “the other solution is cylindrical - ours give less cylindrical which is good” (paraphrased) … But if you can’t control it I’d say that I’d prefer my cylinders staying cylinders and I’ll model less cylindrical if I want them to be less cylindrical?

I love it when people do research on new skinning ideas - but this seems very biased in my opinion, not discussing the potential negative aspects.

The results does look quite good overall, but I really hate it when I have to try to figure out the potentially negative things myself because the author is ignoring, missing or hiding them.