Is PPT resolution dependent?
I know that the LightCache isn’t, but in PPT would a scene at 500pixels and say 5000 subdivisions be EXACTLY the same clarity as a scene at 3000pixels and also with 5000 subdivisions?
The reason i ask is because i have a simple scene that unbelievably is getting to an acceptable (ish) level of noise within 10 minutes at 800pixels wide, with 3500subdivisions of PPT. I thank adaptive tracing for this…
For the final render i’d like to set it to 7000 - 10000 subdivisions, which i approximate should take around 30 minutes - 45 minutes and should give a fairly noiseless image. (Or at least acceptable)
I want to submit it to the render farm so i don’t loose a render node to only one image over night, so i’m just trying to estimate how many subs i will need. If it isn’t resolution dependant, i’ll stick with 7000 - 10000, but if it is, like Irradiance, should you double the amount of subs as the image size doubles? (i.e. like Irradiance, halving the min and max values as the image size doubles)
it does.
i prepared a LC samples calculator for that purpose.
and posted it too, somewhere.
found: -->here<--.
set the pixel size to 1, and then the sampling rate (rough idea of quality)you wish.
It’s especially helpful if you test problematic areas out with a crop render, at full rez and full quality, and use the calculator to adjust it.
Don’t know about the adaptive part of it, sorry about that.
so I thought The actual samples done is like 500x500 = 250´000. If you increase the sample rate to 5000 then we got 5000x5000 = 25´000´000 wich is a 100 fold increase counting actual samples. at the same time the sample indicator is multiplied by ten?
you’re right.
The sample rate does match, in ppt mode, what vray outputs (47.03, not adaptive).
I seem to remember there’s a square root somehwere in the code, and i should check: it’s been a LONG while since i last opened the code.
You can, anyways, do time tests by yourself really quickly by timing crops’ rendertimes.
Rezolution: 1000/1000px , Subdivs: 1000 - Clean image (Y)
Rezolution: x2 - should have Subdivs x2
So if you want a 5K wide image, you would have to multiply x5 the Subdivs, in order to get the same quality.