Sampling affects brightness

[fAEkE]

I know why sampling affects brightness of pixels, but I made a test which in my eyes is kind of interesting. Take a look at the image.



Three tests with different settings as shown beside each render. Now, whats interesting to me is the changes in brightness depending on samples taken. With low settings, you get a brighter image. I guess vray gets less secondary samples from the environment and therefor leans toward the brighter primary rays from the light.

With higher primary rays (DMC max subdivs) and lower secondary samples (light and material subdivs) you get a darker image, but with better quality. Nothing strange.

Now to whats interesting. With even higher samples in the third attempt, you get a brighter image than on the second attempt. Why? Shouldn't this be exactly like B or maybe a bit darker?

I'm confused. As always.

[joconnell]

This is just a theory of course, with lower rays and less samples, the bright samples are overpowering the dark samples, giving you a noisy, bright image. With more samples in the second render, you're getting less bright samples from your materials so there's less bright samples and the anti aliasing is gradually averaging out the bright values. The third render has the most samples and is the most accurate overall, showing that the bright sample solution is actually correct.