Chaos Corona Forum

Chaos Corona for 3ds Max => [Max] Bug Reporting => [Max] Resolved Bugs => Topic started by: FrostKiwi on 2015-02-02, 21:59:27

Title: CoronaAO wrong samples count math?
Post by: FrostKiwi on 2015-02-02, 21:59:27
Hey,
before plastering Mantis with minor stuff that may or may not be wrong:
CoronaAO has the MaxSamples (per pass?) setting. Disregarding AA, 1 Ray per Sample with 1000 passes obviuosly renders 8 times slower than 1000 Samples with only 1 pass with identical Image result (Attachments), however the maxsamples setting in CoronaAO may be incorrect:
setting it to 1 produces 5 Rays/sample instead of 1 per sample, thus 200 Passes with maxsample setting 1, equal 1 Pass with maxsample setting 1000 in identical imagequality and rays cast and NOT 1 pass with maxsample setting 200, like one would assume.

This is a minor nitpick I suppose, yet still an incorrectness with the overall Rays/Sample count.
Lastly, Corona needs to clerify in the Wiki or documentation that although the image is 1:1 the same in 1pass with 1000samples and 1000passes with 1 sample (disregarding AA), it does NOT equal the same rendertime by far. This may or may not be important when balancing Samples and Rays for an Animation.

(eg. I found out, with simple geo and not too complex normalmaps 8passes is enough at 1080p in terms of AA, thus I limit passes to 8 and adjust Lightmultiplier and GI to kill noise, so that Animations render fast, without loss in quality)
Title: Re: CoronaAO wrong samples count math?
Post by: Ondra on 2015-02-02, 22:42:40
the reality is more complicated, there are also other rays, such as the primary, some adaptivity, etc. But yeah, it is internally being capped to be at least 4. So I changed the minimal value in UI
Title: Re: CoronaAO wrong samples count math?
Post by: FrostKiwi on 2015-02-03, 01:09:44
the reality is more complicated, there are also other rays, such as the primary, some adaptivity, etc. But yeah, it is internally being capped to be at least 4. So I changed the minimal value in UI
Before I start another thread to bitch and moan in :D ...
... shouldn't AO Occlude based on NormalMaps and Bump maps aswell?
Normalmaps appear in the shading around geometry corners, yet are ignore as being the source of shading themselves.
See right wall in attachment for extreme example