You may need to do an “incremental” on our IR maps, Actually it’s best to divide the list, combine them and then combine the larger IR maps.
We’ve a system here that does just that. We usually only render every 50th or 30th frame and then add specific frames when coming through doorways or fast turning pans… Your camera would have to REALLY be moving fast to need every other frame.
With the dependencies setting, I’m able to send a complete job at once. The first is the multi-frame “50s” past the renderers the IR maps, the frames are set via the “pickups” box to control specific frame rendering. The second a command line job, that uses a python script to control the merging, in the case of too many maps… ect… The third, and fourth if Strip rending, is the final render.
The light cache is the computed in fly-through and saved out by the box rendering the first frame of the “50’s” job.