I’m pretty sure I know the answer to this, but I just wanted to check, to make sure I haven’t missed anything. In the Maya plugin, is there any way to preserve volume in the fat layer on self collision?
To give a good example of what I mean, say for example you are using the sumo character shown in some of the demos. If I animated an extreme pose, like that sumo touching his toes, then there would be a lot of interpenetration of the belly into the thighs, and probably arms as well. I know I can turn on self collisions on the final skin, and that will create an offset gap, so something like a cloth sim won’t pinch, but that will just offset into the belly, causing it to lose volume. Is there any way to simulate the belly getting squished in a realistic fashion?
I know Maya doesn’t have tetrahedral primitives like Houdini, which makes volume deformations like this difficult, but I thought I would just check if there was maybe a setting or technique I missed.
Thanks for posting your question in here so other users can find it helpful. Find the reply below:
Regarding the volume preservation you are describing, AdnFat is designed to provide that kind of volumetric behaviour. So, for example, when the belly is compressed against the thighs, the fat simulation can preserve volume and allow the tissue to bulge and redistribute rather than simply collapsing inward.
The current limitation is that AdnFat does not yet support self-collisions directly. This is something we are looking into and is on our roadmap.
In the meantime, one approach we would suggest is to approximate those contacts using the fascia self-collisions. On the fascia layer, you can paint the regions where you expect contact, such as the belly, thighs, or arms, and then increase and paint the self-collision thickness to roughly account for the thickness of the fat sitting on top of the fascia.
The idea is that the fascia starts colliding before the underlying surfaces get too close together. The AdnFat layer on top can then still provide the volume-preserving deformation and squashing behaviour, while the fascia collision helps prevent excessive interpenetration.
Finally, self-collisions on the skin can be used to resolve any remaining surface intersections and clean up the final result.
It is a workaround rather than the ideal solution. Ultimately, we would like the AdnFat solver itself to handle self-collisions so that this kind of contact can be solved directly together with its volume-preserving behaviour.
I haven’t had a chance to try it out in the rig yet, but I think putting a thicker collision on parts of the fascia will definitely help. I’m very glad to hear this is on your roadmap, and look forward to seeing what future updates hold.
Just to update this, the thicker collisions on the facia do help some. It isn’t a perfect solution, as it still doesn’t actually preserve volume, and it doesn’t do as good a job from all angles. For example, if you have a fat character who bends over to touch his toes, the thighs will collide with the belly, but the belly won’t really get squeezed by the upper chest, and neither of those collisions will push the belly out to preserve the volume like it should. But it does give a more natural looking effect than just processing the collision on the skin layer.