Tweet
I have an interior scene lit primarily by an HDRI in a dome entering into the room through a number of smallish windows.
I was noticing that from render to render occasionally the overall brightness seemed to be changing, quite noticeably on one material in particular
and started trying to track down the culprit as I was sure that I wasn't changing light/camera/material settings.
So it turns out that the more samples available to the scene the brighter the output from that HDRI dome gets (visible in VrayLighting & VraySpecular elements).
In particular if I set the adaptive amount to 1.0 with a lower number of min. samples, say 8, it renders almost black in these elements on this material (float 0.007)
but changing the adaptive amount to 0.95 the result is much brighter (float 0.04).
Previously I'd just assumed that settings with less samples would produce more noise and not a marked reduction in brightness itself.
I'm guessing (with my limited understanding!) that the brighter version is the more correct one and that the samples are necessary to trace
enough light patches from the sky (which is cloudy) through the small windows onto parts of this material.
Can anyone let me know if they think that's right or if there might be something else at work here?
Also are the influential sample amounts in this situation just those that are applied to the Dome light itself?
I was noticing that from render to render occasionally the overall brightness seemed to be changing, quite noticeably on one material in particular
and started trying to track down the culprit as I was sure that I wasn't changing light/camera/material settings.
So it turns out that the more samples available to the scene the brighter the output from that HDRI dome gets (visible in VrayLighting & VraySpecular elements).
In particular if I set the adaptive amount to 1.0 with a lower number of min. samples, say 8, it renders almost black in these elements on this material (float 0.007)
but changing the adaptive amount to 0.95 the result is much brighter (float 0.04).
Previously I'd just assumed that settings with less samples would produce more noise and not a marked reduction in brightness itself.
I'm guessing (with my limited understanding!) that the brighter version is the more correct one and that the samples are necessary to trace
enough light patches from the sky (which is cloudy) through the small windows onto parts of this material.
Can anyone let me know if they think that's right or if there might be something else at work here?
Also are the influential sample amounts in this situation just those that are applied to the Dome light itself?
