Copy LightCache and IR Maps locally pre-render

[Dave_Wortley]

Might be a bit of a pig to impleament because ideally it requires max to communicate with Backburner in an ideal world but there are work-arounds.

I've noticed significant improvements to rendering timings if i use a pre-render script I wrote that copies the LC and IR maps to the local machine from a network location before rendering then re-paths the maps to local. It seems that all rendering servers fighting to read the same file causes a massive slow down on our network at least.

Would be good if Vray did this internally... My script works well because the blades only have to copy the file to local once per job and it stays there until I send a delete job to the blades, in an ideal world it would know when the job had been terminated/finished and then would auto-delete, but this requries either communicating with backburner or a test to see if the current job is different to the last job, if true, then delete old map, bit of a hack-around but does the job.

[MegaPixel]

It would be great if that little check-box in the network render dialogue that says "Include project bitmaps" could be expanded on to include all project assets, including Proxies and even Precalc content if needed (LC/IR maps). It wouldn't always be very efficient to copy precalc content to every server though, especially for projects involving multiple camera shots using frame by frame ir maps because of moving objects - which would add up to a crap-load of files - Plus depending on which frames the servers participate in, they probably wouldn't use all of that precalc content anyhow. The biggest network performance hit always comes at the beginning during initial project load up which is generally unavoidable, but once a project is loaded into memory, it should stay there until the project is complete. If you factor in the network load to perform your script mentioned above, you are essentially pounding the network just as much by mass distributing all of that content - Depending on the size of the project, this might not result in the performance gain you are looking for - Shrugs