With traditional irrmaps, there was a nifty trick where if you did say a irrmap + lightcache calculation, and saved the irrmap, then you could render the final pass with only the saved irrmap and no LC. This meant faster renders as the LC data was baked into the saved irrmap solution.
I’ve noticed that with the new animated irrmap sequences, this is not the case. Using the same approach will work, but the final render will be incredibly slow if you don’t have a secondary bounce solution turned on. Is this because the LC is not baked into the saved irrmap?
It doesn’t make sense that working from a saved animated irrmap sequence takes so much longer than single frame mode. The irrmap settings are all identical between the different attempts, as are the LC settings. The only thing that changes is whether I work in single frame mode or save to an irrmap sequence.
Bizzare… I just did Irrmap Animation Rendering mode again and it blew through the frame in 15min… making the total time almost the same as single frame mode. Although all my tests yesterday were coming out to 3+ hours for the same configuration. I’m getting very confused here and will have to do some more testing.
Well, it could happen if there is some discrepancy in the scene between the times the irradiance maps were created and the final rendering (e.g. objects added, removed, or some other change).
Right now I’m doing some test with animated irr mode, and it seems that single frame is 3-4 times faster than precalculated irr in animation mode.
No changes were made in the scene, I just precalculated IRR in animation (prepass) mode (too bad all dr nodes cannot be used to help here - only main machine is used to calculate irr) and then switched to animation (rendering) mode - 1 frame for interpolation. And as I said - 3-4 times slower than single frame mode…
Single frame - 26s.
Animation - 3m19s
The only thing that is animated in the scene is the sun. Camera is not moving. There are some translucent objects, but I don’t think that this is the issue here…
2 DR nodes + main machine.
any thoughts?
edit: my further test show, that it’s even 7 and more times slower than single frame mode, even with only one frames being interpolated. I know that animation mode WILL be slower, but why that much?
Third that. I had the same experience the other day.
I’ll check my scene for the “Use light cache for glossy rays” or the “Treat glossy rays as GI rays” but those settings are defaulted to be unchecked and “Only for GI rays” respectively, correct?
In that case, would be best if one of you can send me a scene to vlado@chaosgroup.com for investigation as I have no immediate idea of what might be causing it.
At this point, the scene is behaving as expected and I can’t replicate the problem. I have had it occur on other scenes as well, so I’ll dig something up and get it to you vlado. Thanks for paying attention to this.
We are having the same problem here. We try with two completely different scenes, and the render times with animation mode are massive when comparing with single frame. We run Max9x64b and Vray1.5.
I can’t help with the problem here but I am having a similar problem using a non-animated saved irr. map that’s taking longer to render than a calc’d one wen there’s a lot of proxies. I started a new thread here…http://www.chaosgroup.com/forums/vbulletin/showthread.php?t=41292 Maybe the problem is related.
