bits missing - irr map -animation

[mcvltd]

Having a problem with creating an animation using irr map for primary bounces and lc for secondary.

The lc and irr maps are precalculated on one machine using flythrough and multiframe incremental repectively. Then distributed between the two render nodes (using the same path names of course) and I end up with chunks of render missing. (see image below)



If I use the precalced lc map for rendering of primary and secondary bounces then this problem disappears, so there seems to be something wrong with the precalc'ed irr map. It's really odd because I've used this exact same method successfuly for two previous animations (but using different scenes). I've even tried re-calculating the irr map but with exactly the same problems.

This doesn't seem to happen at all with still images rendered with single frame irr map.

Anyone got any ideas as to what could be up?

Regards,

Rob.

[sv]

only thing I can suggest is to re-calculate both the IR and LC again, to a completely new, clean file.
Also, try Incremental Add to current map, as i think i'm correct in saying that it doesn't clear the last frame from memory, like multiframe incremental, but keeps it in memory and only calculates the new samples it needs.

How often did you calculate a map in multiframe mode - i.e. How many nth frames did you set it to precalculate at? If it was 30, try 20, or if it was 20, try 15 for example. It might have been that the camera couldn't see this bit of the mesh as it was occluded by something, but looking at the image i wouldn't have thought so....

[mcvltd]

Thanks sv, maybe I'll try incremental add. I've recalc'd the maps twice using the current settings but with same results. Nth frame was set to 10 and total frames are 180 - it's just a very slow camera transition from right to left.
Will report back once I've re-calc'ed.

[sv]

do the error's dance around, or are they static in the scene - i.e. specific to one part of a mesh?

[mcvltd]

They're definately static.

[sv]

hmm, well let us know how you get on with a clean re-calculation in incremental add mode...

[vlado]

What are the settings for the irradiance map (in particular, the "Sample lookup" parameter)?

Best regards,
Vlado

[mcvltd]

I used the medium animation pre-set, then changed it over to custom, made the min rate-4, detail enhancement on (set at 2mm) with interpolation of least squares fit and sample look up of overlapping. Auto save is off - I do that manually when it's finished.

[vlado]

You can try changing the "overlapping" setting to the default one ("density based") and see if it helps.

Best regards,
Vlado

[mcvltd]

Ok, will give that a try if the incremental add doesn't work - still got 4.5hrs of calc'ing to go.
My only fear with going to density based is that, for me, it usually means a shed load of GI splotches. But I haven't tried it for some time though. Will report back in a few hours.

[vlado]

Well, the black parts will appear with the "overlapped" setting if there are too low samples in the map, or no suitable samples in that area; since you are getting splotches with the density-based one as well, that might mean that the GI solution is not very good for this particular camera angle - so the incremental mode should help.

Best regards,
Vlado

[mcvltd]

That's interesting, because this is the only scene that I've tried to light with just the vray sun and nothing else. The previous scenes had some direct lighting.
So I guess I'm asking a bit too much of my current irr map settings.
What would be the best way to increase the sampling? hsph subdivs/interp samples or a higher min/max rate?

[vlado]

Higher min/max rate should fix it (what are these now?)

Best regards,
Vlado

[mcvltd]

min-4 max-1

[mcvltd]

The incrmental add didn't make any difference, but raising the min rate to -2 seems to have done the trick! Thanks for your help guys!