[deflix]

Ive noticed that Maxwell becomes increasingly inefficient at rendering when increasing the resolution of the final image - this doesnt seem proportional. Also when utilizing multilight the memory requirements quickly become prohibitive. Has anyone overcome this by rendering 4 seperate quadrants for example (regions)....and would there be a working method for this approach. Also is a cooperative render really using ALL the nodes power to full capacity -- from my tests so far im not convinced; it seems to me that the final SL achieved with a cooperative render isnt a proportional improvement over a single node render. WIP of scene in question below...


Thanks

[Bubbaloo]

Ive noticed that Maxwell becomes increasingly inefficient at rendering when increasing the resolution of the final image

Really? I've noticed quite the opposite. It takes longer to reach a certain S.L. but lower S.L. at high resolution looks better than higher S.L. at lower resolution.

it seems to me that the final SL achieved with a cooperative render isnt a proportional improvement over a single node render

It will seem that way because Sampling Level is exponential. The higher the S.L., the more time it will take to reach the next S.L.
I have found that to reach the next S.L., it will take half the time of the render so far. So if you have rendered for 20 minutes to reach S.L. 10, it will take 10 minutes to reach S.L. 11.

[deflix]

thanks for the response.

when cooperative rendering for example why do all nodes show the same sl?
Should it not be the case that they are all contributing to the SL or am I missing something.?

[Bubbaloo]

They will show their individual s.l. reached until you stop the render, then it will merge them to the accumulated s.l.