Area light noise on the side

[isolog]

Area light noise on the side is difficult to remove
This is the result of 3000 samples from the PT engine

oc

This is the result of rs double brute force automatic sampling 0.01

rs

So is there any way to render noise cleanly without denoise?

[elsksa]

Hi,

While Octane is doesn't have a ton of settings, there are too many to check. It's easier to point out to this Kernel Settings Guide and this Octane Light Guide. It covers nearly everything one needs to know.

However, I can share the following: that's already too much samples.
Did you set the diffuse material of the light to pure black? It's white-ish by default and this is to avoid at all cost.

I invite you to consult the two pages and double-check everything, then report back with screenshots of the relevant settings.

[isolog]

Additional Info:

[frankmci]

Reducing the GI Clamp to something that fits the scale of your scene should help a lot. I doubt I ever use a setting of more than 10, but it defaults to 1000000. I generally start at 1, then tweak from there, if necessary. This is Path Tracing with 300 samples.

[aoktar]

Anyway it looks like a weakness on Light rays which are blocked by self at zero angle between faces. You may export ORBX and post in standalone topic

[elsksa]

Reducing the GI Clamp to something that fits the scale of your scene should help a lot. I doubt I ever use a setting of more than 10, but it defaults to 1000000. I generally start at 1, then tweak from there, if necessary. This is Path Tracing with 300 samples.

That's exceedingly low. Although it wouldn't make a drastic difference in a scene that has weak lights, among some other factors.

Gif demo from this page.

[frankmci]

That's exceedingly low. Although it wouldn't make a drastic difference in a scene that has weak lights, among some other factors.

Hmm. I'm curious to know your experience with GI Clamp. Here's a quickie mock-up of your example, using an Octane Daylight as the only light source. Yes, it's certainly too noisy at this point with only 300 samples and a diffuse depth of 6, but it's a pretty decent starting point. Using higher GI clamp doesn't improve anything. I wonder what we are doing differently. I'll have to read that page you linked to at lunch.

[elsksa]

"The Candle & The Sun".

The "scene" doesn't have enough "energy", yielding in barely any dissimilarity. A stronger light intensity, especially for the sun, will result in a more plausible result and such difference as seen on my gif.

[isolog]

Anyway it looks like a weakness on Light rays which are blocked by self at zero angle between faces. You may export ORBX and post in standalone topic

Thanks, I will.