My setup: XP32 with Max 8 and Vray 1.5 SP1 and 4gig memory and Quad Core cpu
Make scene with above setup but rendered in 1.5RC3. Ok renders this way.
LightCache pass (1200 samples) Irradiance Map passes (3 of these) rendering
takes about 3 hours.
Make scene with above setup in SP1:
Light Cache Pass OK. Irradiance Map Passes (6 of these) crashes on Pass 5. Memory allocation error. Look at memory approx 75% used. Max Crashes before Vray. Hit Close and Max exits.
Tried: reducing to vrmesh. Scaling down texture maps. Using dynamic memory limit. Only thing that worked was reducing res of render from 1000x4400 to max height pixels of 3200. Rendered ok then. Took 10 different combinations before it would render. But It rendered at full res with RC3 on the first go.
The smaller render took 5 hours to complete.
Ok I uninstall SP1 untill I hear, because same thing happens every time i go for BIG render.
Raj
For large renders, it is better to use the V-Ray frame buffer and render directly to a .vrimg or .exr file, with the “Use memory frame buffer” option disabled to save RAM.
Other than that, there isn’t much else that can be done, other than going to 64-bit 3ds Max and V-Ray.
I tried the .exr option already in the frame buffer. (is there a way to extract the renderelements layers (Apart from buying Nuke) cause CS3 doesnt seem to see them, I use Render ID a lot.).
The difference is that RC3 rendered the large image ok, but SP1 didn’t
I’m trying the same render again to see what happens, this time at full res in RC3.
Also why did RC3 do 3 irradiance map passes (medium quality preset) and SP1 elect to do 6?
thanks for the quick reply too btw.
Finally (sorry for the annoying questions) Is there a render element like RenderID that takes into account transparency as well? IE if a renderdered sphere of the world with a tranparency map has render ID - the whole sphere will be a solid colour in render ID rather than just the opaque bits. I would like to be able to layer the image in Phototoshop, and have the objects seen through the transparent part of the sphere on a separate layer (so I can insert text in between for instance). I guess I’m asking for a separate alpha channel for every object combined into a randomly-coloured-by-object single layer.
Is that too much to ask? It certainly is too difficult to say.