Setting Max Num Consec Reject To 0

[nakile]

I decided to run a few test scenes to see what would happen if I did this.

The following were both done in about two hours on a Pentium 4 running at 2.4 Ghz with Max Num Consec Reject at 0.





From what I understand, this is about as bias as Indigo can get, correct? Even so, those images look good for running a few hours, I think.

I guess I'll have to run those scenes again for the same amount of time, but at 1000.

[Kram1032]

yes, you're right and yes, you should

[CTZn]

I hope the <halt_samples> tag will be available soon

It's much more relevant than a <halt_time> for such tests !

[nakile]

Strange... I ran the first scene for only an hour with it set on 1000 and came up with this:

[Deus]

Yeah. Really strange. The real question is where did that monkey head that was in the first image go? The real mystery is begining to be uncovered.

[Kram1032]

that really shows the effect of max_num_consec in an impressive way

Quality gets much better, but noise disappears slower.
Therefore, shadowed regions stay extremely noisy.

-> Indirectly lit scenes and scenes with caustics : high max_num_consec