Just curious about the the ram usage on max-files? Are the settings of vray more important or the poly amount of my model to save precious amount of ram? I know both eat up lots of ram if set or build with excessive use of numbers 
thx in advance
ralf
Usually bitmap textures are the biggest RAM killer, though procedural materials tax the processor a lot more. Geometry is the second culprit.
Displacement is a killer as well.
adaptive sub AA can be a killer too, adaptive qmc is much lighter.
Opacity maps - especially if your trees leaves are opacity mapped.
The render penalty for using opactiy mapped leaves..etc.. can be greatly reduced by turning the filtering in the bitmap properties for the opactiy map to OFF
Would that cause a more aliased result or flickering when animated ?
Nope. Not that I have noticed.
out of interest, when you say bitmap textures are the biggest RAM killer, what the best way to reduce this? will using more aggressive compression in a jpeg help or does it work the same ways real-time 3d does whereby the its the physical pixel size that makes the difference not the file size and does keeping pixel sizes to a power of 2 ie 512x512 or 1028x1028 etc.. have any effect of ram usage?
from what I’ve heard, max will make your image into a power of 2 memory wise. So your better off using that to your advantage.
More compression on a JPG will make zero difference. The only difference will be obtained from a smaller resolution OR using an image format that uses indexed color such as GIF
your serious? i always thought the power of two thing was a myth
i swear i read that somewheres. I wish I remembered where though. I believe max only supports power of 2 textures so it converts it at rendertime. I always hoped I was wrong on that though.
uncompressed image formats are lighter on memory than heavily compressed ones.
Vray AFAIK reads in the image/texture pixel by pixel so image compression should not make any difference.
power od 2
So when you are talking about the power of 2 you mean 1024x1024 and so on right?
cheers
ralf
good question, does it need to be square AND power of 2, or just that each side should independantly a power of 2.
for instance, 1024x1024 or 1024x512?
When I was in school using Maya they told me that you should use square maps that were even. Of course, I knew power of 2 was more efficient, but I didn’t know how much.
Also, how much does that really effect the memory/speed?
To be honest i am not sure if POT (Power Of Two) textures still get you any significant gain those days. It originated in weak hardware actually. The reason why they are/were so much better is that powers of two enable a lot of neat programming tricks that dont work otherwise. But these are mainly processing advantages, not memory advantages. An example would be SHR and SHL instructions. That means simply shifting the bytes to the left or right. Mind you that computers work binary. imagine decimal 2. As binary that is written as 10. Shift to the left and you get 100. Wich corresponds to 4 in Decimal representation. Another shift and you get 8 etc.
So instead of performing a multiplication you simply shift bytes wich is a lot faster of course. Considering the amount of other calclualations done during rendering i doubt that POT textures return any noticable gain…But i might be off there. Mind you that they dont have to be Square. Square Textures do provide additional value in the original idea, as you can ignore either U or V in all calculations as they are the same…But as Textures are hardly ever used in their original format but have to be either scaled up or down for almost every damn pixel (stupid perspective distortion 
) that doesnt really get you anywhere i guess…
Regards,
Thorsten
actually I guess just square. Power of 2 might indeed just be more efficient.
Another thing
anybody else had problems with insufficient RAM when working with the Vrayblend mtl? I can’t get my 2gb machine to stop crashing when using more than two coats in the Vrayblend mtl. :evil:
Am I missing something? 