I am not sure if i am doing this the right way. I’ll describe my workflow:
I have set up a scene, and i have to render some 6 cameras ranging in sequences from 200 frames to 400.
With some slow moving cameras, i’ve set it up to render the irradiance map every 50 or even 75 frames (in case of a very slow camera pan).
The faster moving cameras, which change angles and sporadically include trees and other stuff in the foreground, i’ve manually selected frames where it seemed proper to render the irradiance map.
Now comes the part i’m not sure if i’m doing the right way.
Using the imap control script i send each sequence to be rendered through the back burner. At 900x506 i am rendering at “high-animation” preset.
The sizes of the irradiance maps range from 15 megs to 130. Then, i merge the maps manually (i couldnt get the other script to work 
) which is a tedious task.
One such map, in the end had over 600mb, and it took a while to load.
At this point, i setup the scene to load the imap from c:\ since thats the only drive that my render slaves have, and manually copied the imaps to each slave.
Now, since it seemed that loading a 600mb imap takes a while, i decided to combine them into sequences. So a imap that i rendered at every 20th frame would be merged into sequences that range from 0-40,40-80,80-120 and so on…
It is pretty time consuming to merge the imaps and to copy them to slaves (something not set up right network wise), but it did the trick.
Is this pretty much how it should be done?
It took me a while to merge all the irradiance maps, copy them to the slaves, and then create back burner jobs for each sequence, because i had to manually type in the imap path for each sequence, choose a number of frames and so on.
The GI is flicker free though, and irradiance maps calculate on average for about an hour, allowing me to pump up the AA settings a bit.
But in the end, i had to leave an interior animatic to be rendered on a single machine using incremental add, since there was something wrong with the pre-rendered map which was producing a ton of artifacts.
If i were to render with “incremental add”, compared to the multiframe, would it render as fast?
For examaple, i have a sequence of 20 frames, and i prerender the first and the last using multiframe incremental, and that takes an hour for each. Incremental add would only render the first one for an hour (with the same settings) but then each next one would render for a fraction of the time summing up to two hours as well?
What would be the advantage of one over the other. Multiframe incremental seems to work well with the back burner, but i could also break the animation into parts and have them render single slaves using incremental add. Would this then be more prone to flickering and GI artifacts?
