ChrisCousins

Hi - hope someone can help. I've got some animated foliage (cactus spikes) that I want to scatter over a large area. There's 16 variants, and approx 5,000 spikes.
I want to control the animation with a falloff, so the spikes open and close in waves.

First thought - Alembic Delayed Load! Perfect. Variant attribute is saved nicely, copy to points works great - and using setprimintrinsic(0, "abcframe", @primnum, @retime, "set") to control time. However, this causes glitching and error issues - there's aggressive caching and the instances aren't responding correctly.

So... packed disk sequences. Great - using setprimintrinsic(0, "index", @primnum, targetFrame); I can control the frame with an attribute.
However, when loading the geo is treated as one object, and I can't use the variant attribute. It places all 16 animations on each of the 5000 points.

So the question - is there a way to instance packed disk sequences respecting a variant attribute, and allow for control of time? Or should I save the 16 variations as individual sequences and load them that way?

Fixed! By watching the MPM masterclass, which is super-helpful by the way.
The missing checkbox was this one, which watches out for particles creeping through the collider, exactly what I wanted.

Hi - any tips on keeping MPM solver from messing up collisions? I've got some viscous/sticky fluids inside a sealed collider, using the vdb/invert option in the MPM collider node. The simulation always has leaks/errors, it looks like the sim is lagging the animation and this allows the edge particles to escape. See vid attached

Things I've tried:
1. substeps - min/max up to 200/20000
2. CFL - down to 0.01
3. looser/denser meshing for the collider
4. changing collision type to animated deforming (instead of animated/rigid)

Thanks for any advice! Chris