OTOY Forums

hey people from octane. your renderer is super mindblowing and im addictively working with it every day! thank you for this amazing piece of software!
looking for the 1000th time at the samples rolling in and seeing that brighter parts of the image (more acessible for light) would clear up faster than darker areas (less acessible for light) i was wondering: does a 1000maxspp setting mean every pixel will recieve 1000samples? or does octane sense that noise is completely vanished and stops clearing up the image in these bright/directly litten parts?

i hope i make myself understandable..

nerd question as i said:)))

br.
mo

momade wrote:hey people from octane. your renderer is super mindblowing and im addictively working with it every day! thank you for this amazing piece of software!
looking for the 1000th time at the samples rolling in and seeing that brighter parts of the image (more acessible for light) would clear up faster than darker areas (less acessible for light) i was wondering: does a 1000maxspp setting mean every pixel will recieve 1000samples? or does octane sense that noise is completely vanished and stops clearing up the image in these bright/directly litten parts?

i hope i make myself understandable..

nerd question as i said:)))

br.
mo

If you are using the path tracing and direct lighting kernels, the sample distribution is uniformly, i.e. every pixel receives the same amount of samples. PMC on the other hand concentrates its sampling effort onto areas with high light contributions (i.e. bright areas).

There are two reasons why darker areas clear up more slowly:

1) Dark means often that no direct (i.e. not shadowed) light is reaching those areas, which means that all light contribution comes from indirect sources, which are often hard to find. This means only a small percentage of the paths contribute light resulting in more noise.

2) Dark also often means dark surfaces and it's more likely that a path is terminated earlier, because the noise is less visible on dark surfaces. You can control this early termination by changing the "path termination power" in the kernel settings. This parameter specifies the maximum threshold up to which a path may be terminated earlier. Reducing it will decrease the probability that paths are terminated earlier (resulting in less noise per sample), but also decreasing the number of samples/second.

thank you for this precise answer!
very apreciated!

br.
mo