We have scenes where we are rendering a lot of sphere particles in vray, but are limited by subdivisions/memory. What is the deal with particle instancing, sphere particles and Vray that eats so much ram?
We also have instanced geo particles that are severely limited in how many we can render due to going well over the 12 gigs of ram we have.
basic stats:
50,000 sphere particles with sphereSubd set to 16 (we need more actually… we get quite close). runs out of ram.
I would think that:
1. a sphere wouldn’t take up so much ram, even with a million of them.
2. geo instancing should be more efficient. Is there a way for vray to take any instanced geo and automatically convert to a vrmesh?
On the same subject, if we have a bird, instanced on 1000 particles :
-Does VRay allocates memory for 1 or 1000 birds ?
-Does VRay allocates memory for a bird when the render reaches its bounding box (this method reduces the memory needed to render), or at the beginning of the render ?
So to get back to the original issue; it looks like for “sphere” particles, V-Ray does not instance the the spheres for the moment. I’ve made a note to correct this (by using perfect raytraced spheres), but in the meantime you can just use an instancer node.
Yes, that seems correct. My issues with instanced Geo were more related to the shader complexity than number of instanced particles. Currently, sphere particle types don’t seem to “instance” and take up large amounts of ram. As soon as I use an instanced sphere set up manually, my ram drops down. Thanks for the info. Hope to see it fixed, but in the meantime, we’ll do what is suggested, and instance our own geo. Wish I had known this a month ago;)
As a side note… I decided to try instancing a million low res geo particles with a default shader. It pushed close to our 12gig ram limit, but did a good job.