Uniform Volume rendering of particles - inexplicable render time spike

[danw]

I'm rendering foam for a water sim, with the "Uniform Volume" property applied, and a basic Volume Cloud shader (with the shader density matched to the Volume Density property)

I'm using the Raytracing engine ( - as this doesn't need to generate micro-poly geometry for the Sphere primitives the particles are rendered as, and renders quicker with lower memory usage)

This works fantastically, even with 5million+ particles... except in two or three buckets of the render, where it'll hang on them for 15-30 minutes... the whole rest of the render will rocket through in under 2.

I've switched off every other element of the render, it's definitely the uniform volume particles causing it.

The thing I really don't understand is that it's not linked in any way to the density of particles... it renders the densest areas almost as quickly as the lightest... the couple of buckets that get stuck are in relatively thin areas with only a few (well, only a few hundred) particles.

I've tried tweaking just about everything I can think of:

Stochastic Transparency - doesn't have any effect on Uniform Volumes, as it's not stepping through an actual voxel-volume.

Volume Step Size - same deal, doesn't matter what it's set to, it doesn't affect Uniform Volumes.

Opacity Limit - thought this might have some effect... maybe the rays would give up after a certain density was reached, but no effect at all.

Raytracing Bias - dunno if this applies in any way to uniform volumes, but it didn't have any effect either.

I'm using a VOPSOP that uses point-cloud functions to flag any particles completely-contained-inside any other particles for deletion, which thins out the number of particles by about half, and ensures there isn't too much overlapping particles going on. This actually has an effect, before I was getting about 6 stuck buckets, now it's down to two serious ones.

There's no displacement, no reflection/refraction rays on the volume itself (water surface is reflecting/refracting, but it's not reflecting/refracting the foam, and it does this even without the water surface in the render)

I'm a bit stuck at this point... can anyone familiar with Uniform Volumes suggest anything I haven't thought of?

Edited March 4, 2013 by danw

[sam.h]

It does seem like the problem is lots of particles stacked on top of each other -- especially since your filtering reduces the amount of stuck buckets...

Have you tried reducing the bucket size?

[dobin]

as you're not using any raytracing on the particles rendering the geometry as points should fix that problem.

[lijunhong]

Hi

foam simulation was very good! Can you tell me how to make it? I hope you can give me some suggestions and related files! Thanks!

[eko]

Nice foam! Can you share some tipps! Thanks!

[djwarder]

+1

Would love to know how to get foam like this! I've tried the usual settings in H12.5 (and Naiad in the past) and never get anything looking this good! Any tips?

[Owl]

+1

Count me in

[Cuckon]

+1

Best CG foam I've ever seen.

[slamfunk]

+1 sick foam brah!