So, I thought I’d be putting H15 through it’s passes a little too look at performance improvements … In my experience, reusing “old” workflows – that have proven to be able to create great results many times over – often yields fun surprises, so I went back to the old POP-Solver (NonDOPs) and had a play.
WOW!!!! I stopped at 4.24BILLION particles taking around 5gb RAM!!! Coulda done a multitude of that on my laptop easily) Nice one SideFX!
Close to 50millionen particles (with a bunch of layered noises and schtuff), 360frames, old single CPU i7 laptop with 16gb RAM (never used more than 6-7gb) … done after 3hrs!!!
Here’s a couple testrenders:
Trail of pointemitter with a curlnoise VOPSOP to randomize it’s path … plus simple one layered curl noise pop on trail articles
Trail of same source but sphere particle emitter instead … layered noises in POPs
Closeup with large spheres Closeup with smaller spheres
And some quick rendertests with PBR and the new Disney shader:
200millionen particles (delayed load) ca.3gb RAM @2min/frame – again old single CPU i7 laptop
250millionen particles (2 seperate delayed loads) ca.3.8gb RAM @3min/frame – again old single CPU i7 laptop
4.24BILLION particles (1 delayed load re-used and altered) ca.5gb RAM