Stats about scene. Using 300,000 hairs created using Shave 8 in Maya 2014, rendering in buffer mode. 1 vray area light. im and lc on med (default) settings. Render is 960X540 with hair taking <1/4 of the image space. Bucket size is set to 32. Hair and geo set to static. Aliasing set high to 1 and 32 using adaptive dmc. Vray v. 2.40.01
When I am rendering 300k hairs in Maya using default settings of im and lc and 1 vray area light, I need to set my dynamic memory to the max I can (60 gigs) to be able to get the best performance, but even at such a high number, I still run out of memory and the rendering comes to a crawl during the actual rendering part after the gi has been calculated.
If I don't use gi, my memory use still gets up to about 50 gigs. It doesn't seem to be releasing memory as buckets finish and move on. If render is cancelled, it takes a long time to release the 50 gigs of memory.
If I switch to Geometry mode, it uses about half the memory (20 gigs no gi, 33 gigs with gi), but the look is far, far inferior (in this case at least) to that which is produced by the buffer mode. It seems that the geometry mode only uses less memory because it renders about twice as fast as when using the buffer mode using the same render settings... therefore, it uses half the memory. There is a bunch of splotchiness created by using geo mode, so I would most likely have to increase hair to solve this which would increase the render close to that of the buffer mode.
Hair prims is out of the question because of instability issues.
Is there something I am doing wrong? What are your suggestions? This is only the fraction of the hair count I will need to be rendering (eventually in the millions). So it would be great to hear any thoughts or ideas.
Would messing with raycaster params help? Is there a way to force vray to dump memory as it goes along? I'll look into optimizing render settings to get the time, and hopefully memory, down... but it seems I will need to do more to be able to get this working.
Thanks for all of the help!
When I am rendering 300k hairs in Maya using default settings of im and lc and 1 vray area light, I need to set my dynamic memory to the max I can (60 gigs) to be able to get the best performance, but even at such a high number, I still run out of memory and the rendering comes to a crawl during the actual rendering part after the gi has been calculated.
If I don't use gi, my memory use still gets up to about 50 gigs. It doesn't seem to be releasing memory as buckets finish and move on. If render is cancelled, it takes a long time to release the 50 gigs of memory.
If I switch to Geometry mode, it uses about half the memory (20 gigs no gi, 33 gigs with gi), but the look is far, far inferior (in this case at least) to that which is produced by the buffer mode. It seems that the geometry mode only uses less memory because it renders about twice as fast as when using the buffer mode using the same render settings... therefore, it uses half the memory. There is a bunch of splotchiness created by using geo mode, so I would most likely have to increase hair to solve this which would increase the render close to that of the buffer mode.
Hair prims is out of the question because of instability issues.
Is there something I am doing wrong? What are your suggestions? This is only the fraction of the hair count I will need to be rendering (eventually in the millions). So it would be great to hear any thoughts or ideas.
Would messing with raycaster params help? Is there a way to force vray to dump memory as it goes along? I'll look into optimizing render settings to get the time, and hopefully memory, down... but it seems I will need to do more to be able to get this working.
Thanks for all of the help!
Comment