Page 1 of 1

Does Maxwell give a calculation bias to some emittors?

Posted: Mon Sep 10, 2012 4:29 pm
by bograt
Hi All,
This is something I have always wondered but in my current case may be critical..
I have a scene with a lot of lights (about 200) (all ies files, 2 emission materials),
the lights are in 2 groups, one has about 170, the other has 30... the group of 30 seems to render a lot slower than the 170 and leaves very noisy grey patches while the other lights seem well 'exposed'. is there a priority to the greater number of emitters?
Does the IES sphere size make a difference to the render speed?
Thanks guys

Re: Does Maxwell give a calculation bias to some emittors?

Posted: Mon Sep 10, 2012 4:57 pm
by David Solito
Perhaps, you could make 2 render passes with one group off.
Maybe it could help

Re: Does Maxwell give a calculation bias to some emittors?

Posted: Mon Sep 10, 2012 6:41 pm
by tom
Bograt, it's not something we can call "bias" because, it will surely converge to the correct unbiased solution in the end. As you say, depending on emitter powers, some of them may converge/sample faster and the others slower. This is a strategy/optimization for helping most common scenarios look visually better in early samples. As a nice workaround you could set all emitter values balanced/equal and start a Multilight render, then diminish the weak emitters during or after the render. That'd bypass the optimization and force Maxwell to sample all emitters equally.

Re: Does Maxwell give a calculation bias to some emittors?

Posted: Mon Sep 10, 2012 6:57 pm
by bograt
Thanks David, your suggestion works well... in fact it actually seems to be an improvement on the light mixer to some extent, apart from the fact that I need to do the blending in photoshop... but I guess with linear dodge blending mode it is the same result right?
Thanks Tom, got ya. I guess I missed the fact that the lumen output of the Ies files were so different.
Looking back I can see how I could have saved a lot of time on past renders, is it just the case that on render start all the emitters should be more or less on the desired brightness?
Thanks

Re: Does Maxwell give a calculation bias to some emittors?

Posted: Mon Sep 10, 2012 7:39 pm
by bograt
Tom, I have tried what you suggested but I must have misunderstood because I cannot seem achieve the desired results... I have tried two settings, one with both the ies emitters at the desired brightness, and one with the ies files multiplied according to their frequency throughout the scene e.g 150 lights 3 times dimmer than the 50 lights.
As you mentioned it is at low sl levels that this effect is worse but the nature of my work means very often I have to provide quick previews.
I was wondering if you could help me understand exactly how maxwell fires out the 'photons', I have always assumed that maxwell emitters simply blast out rays and the engine gathers samples per pixel and through an anti-aliasing like effect the accuracy of pixel colour slowly refines.. it this correct?
how does Maxwell 'optimize' the emittors, is it based on emittor output? or reflected surface area?
I am also curios about the relevance of the sphere for the ies emitters, Once using vray for rhino I faked an Ies using a sphere with a transparency mapped material.. is this similar in Maxwell or the sphere just a kind of placeholder to control the position?
Sorry to bombard you with questions,
thanks

Jules

Re: Does Maxwell give a calculation bias to some emittors?

Posted: Tue Sep 11, 2012 1:06 pm
by tom
Right, it's primarily but not solely tied to emitter intensities. Can you show screenshots of your lighting setup so we can have a better idea about how the luminaries are positioned along the scene?