I would generally have a second pointcloud for each one, called "pointcloudName_CACHED", and load the ice cache on it. Then on this cached cloud add an ice tree for your size and instanced geometry. Seems a bit more work, but works well.
One thing I notice with Arnold, is that it doesn't like negative size particles. So if you are usingrandomizing the size, make sure the particle are not going negative. On 13 November 2014 13:19, Morten Bartholdy <[email protected]> wrote: > Hi Cristobal > > > Arnold SiToA and each pointcloud lives under its own model. I have not > saved the cache to the mixer, as it is faster to keep the pointclouds ready > to simulate by unticking the lock in the Cache Manager. > > > I am rendering instanced geometry btw if that matters - it is easier > to quickly adjust particle size without affecting the simulation this way. > > > Morten > > > > > > Den 13. november 2014 kl. 13:43 skrev Cristobal Infante <[email protected]>: > > > Hi Morten, > > it could be so many things, are you rendering with Arnold or Mental Ray? > > I've encountered similar problems when pointclouds that live under the > same model and share the mixer. Is this your case? > > On 13 November 2014 12:35, Morten Bartholdy < [email protected] > wrote: > > > > > I have a number of pointclouds with a lot of particles in them which I > cache before rendering on the farm, using the Cache Manager Simulation tab > => Caching Files => Cache on Playback and then Lock Playback Cache. > > > > Usually this method works fine, but with my current setup it comes up with > some funky playback of some of the cached pointclouds and unfortunately > this reflects in errors in the rendering of some of my passes. Sometimes > when I load a scene with cached pointclouds, the two front pointclouds will > play back sort of as expected, but with a global position offset, and > fthese two pointclouds fail to render while the rest (which are identical > with another seed and startframe) render fine. > > > > I am rendering particles in one pass and shadows in another for > compositing in to a live plate, and the weird things is while the various > particle passes render fine for all frames, the shadowpass at some point > during the shot suddenly miss some of the cached pointclouds - specifically > the two closest to the camera. I have no idea if this is a factor, but > would like to hear if there are some do's and dont's regarding this > workflow so I can get things done without recaching and rerendering a lot > to get through this shot. > > > > Morten > > > > > >
