Coffee Grind - Leaking particles

[pmvis]

Hello all,


I'm trying to recreate an effect of coffee being grinded similar to the reference attached.


I'm using a force to spin the beans that are then broken up using a voronoi fracture, which is activated with a object test moving from bottom to top.


I'm having an issue that when the particles are broken with the voronoi fracture they start to leak through the container although it's closed. This is giving me problems at render time as the geometry won't load by the renderer (Corona Renderer) I'm using.


Also does anyone have an idea of how to break the beans into fine powder. I thought of adding a series of voronoi fracture events to break down beans finer but it's very slow to calculate.


Thanks for the help.

[tyFlow]

Trying to Voronoi Fracture particles down into dust is going to be ridiculously slow. Same for PhysX. I'd recommend a hybrid approach where you keep big chunks PhysX and when they reach a certain small size you convert to grains with Particle Physics. The downside is that Particle Physics and PhysX don't have 2-way interactions....but with enough motion (assuming this isn't a slowmo shot), the intersections should not be noticeable. As for Voro Fractures...switch to just doing a spawned particle approach (Spawn operator - "voxels inside shape" mode) once they reach a certain small size...that'll be much faster than Voro Fracturing tiny particles. The idea is to take your small particles, fill them up with smaller particles and then delete the parent, instead of performing a slow slicing operation hundreds of times on the parent.

As for leaking particles...not sure why that would prevent Corona from rendering....but there are multiple approaches you can take: 1) Increase sim timesteps and/or PhysX steps. 2) Enable PhysX CCD. 3) Use a Surface Test to test if particles leave your container...if they do - delete them. A combination of all three may give you the best results.

One final note about the Surface Test - currently it integrates velocities before testing, which in your case (using it to test PhysX particles) may lead to particles being deleted even though they don't appear to leave the volume (because the velocity used in the Surface Test will be different than the velocity changed by the PhysX solver at the end of the time step). I am going to make the integration optional in the next build, so that this issue won't come up and Surface Test will be more reliable for situations like this (where you need to test whether PhysX particles are inside/outside a volume).

[pmvis]

Thanks very much for the fast and detailed response. I appreciate it!

I've attached the adjusted flow below but I'm unsure if I understood the set up correctly. Also with the spawn operator I'm unsure how I delete the parent over time?

The leaking particles has improved a lot with your suggestions apart from the last event where I switched to collision as it's faster than Phyx collision. I actually previously tried the surface test method to delete the particles outside the volume but like you said it ended up deleting most of the particles in the volume.

[tyFlow]

Make sure you update to the latest build of tyFlow (v1.017). You're a few builds behind and "delete parent" in the spawn operator being greyed out was a bug that was fixed.

[pmvis]

Thanks for your help on this.

I still can't get it to work and the particles still leak quite significantly. I'll have to come back to this ideas at a later date.