You had only noise level as stop condition? Can you please show the actual number of passes for frame 90 and 105?
Yes... I verified that the frames, before and after finished at 20 passes.
If you have no render stamp element configured you may still find it in the logs. Even if cam movement is very subtile it´s possible (and likely) that your reach those 10% noise at different pass numbers, thus at different render times.
This scenario wouldn't account for the fact that the frame EXACTLY before this restart was 60 minutes, and the frame EXACTLY after I stopped the render started back up again at 50 minutes plus. Further, I performed another test render on the shorter frame times, and it rendered at 60 minutes. Interestingly, however, the frames are IDENTICAL. So the frame rendered, same passes, same elements; just 3x faster.
This render time was SPECIFICALLY attributable to ONE of 9 start/stop render sessions between October 27 and today. EVERY other session was consistent. THis one was not.
I think if we could isolate the event that TRIPLED my render time, possibly everyone could benefit from this. I simply know that from my end, nothing changed.
To further add to the head-scratching, I save my project file after each render session, so had I inadvertently changed something in this session, It would have manifested itself in the next render session as well.
Something happened here that SIGNIFICANTLY increased my render times with the SAME renders.
Good Luck
[/quote]
[/quote]