Isaac Criddle
isaacc
About Me
Isaac works for SideFX as a Film & Television production consulting intern. He has a dual background in FX animation and data science.
専門知識
VFX Artist
Houdini Skills
Availability
I am currently an Intern at SideFX!
Recent Forum Posts
how do I calm down bubbling and jittering and vellum 2026年10月6日14:46
Simulating many layers of cloth is an infamously challenging problem in physics-based animation, and is an active area of research both inside SideFX and at large. The case of dropping a stack of stiff paper is especially tricky because of the way forces accumulate as they move up through the layers. Think about what happens when you drop a baseball and a basketball together with the baseball on top--the baseball gets flung much higher than it started as it catches the bounce energy of the basketball underneath.
Vellum jitters if the physics at a given substep haven't converged. Some things that can help in general:
1. Increase your substeps. This means the sheets will travel less distance between each physics calculation.
2. Increase your collision iterations. This will let the solver have more time to converge at a given substep. If you have a lot of layers of collisions, the number of collision iterations you need will increase accordingly, since the "shock" of collision will need to travel from the collider all the way up to the top sheet.
3. Increase smoothing iterations--this will smooth the error throughout so that bad triangles or bad convergence is less noticeable.
4. There is an example on the content library that removes jitter using Principal Component Analysis as a nice hack. I've attached a very simple implementation of that on your stack example. Check out this article for more: https://www.sidefx.com/tutorials/pca-dejitter/ [www.sidefx.com]
If you need to do a large stack of paper in production, it's generally best not to simulate all of the papers colliding at once, possibly breaking the simulation up into layers, mixing with upstream animation, and/or doing post-processing in SOPs, similarly to your approach here.
Vellum jitters if the physics at a given substep haven't converged. Some things that can help in general:
1. Increase your substeps. This means the sheets will travel less distance between each physics calculation.
2. Increase your collision iterations. This will let the solver have more time to converge at a given substep. If you have a lot of layers of collisions, the number of collision iterations you need will increase accordingly, since the "shock" of collision will need to travel from the collider all the way up to the top sheet.
3. Increase smoothing iterations--this will smooth the error throughout so that bad triangles or bad convergence is less noticeable.
4. There is an example on the content library that removes jitter using Principal Component Analysis as a nice hack. I've attached a very simple implementation of that on your stack example. Check out this article for more: https://www.sidefx.com/tutorials/pca-dejitter/ [www.sidefx.com]
If you need to do a large stack of paper in production, it's generally best not to simulate all of the papers colliding at once, possibly breaking the simulation up into layers, mixing with upstream animation, and/or doing post-processing in SOPs, similarly to your approach here.