Final Gather Rebuild?

How does final Gather Rebuild work?? Does it work like when you re-use disk based D-maps for depth map shadows??

I am about to atempt a sort of architechual flythrough in mental ray with nothing moving but the camera. Should I check my final gather rebuild. If not what do I put in the final gather file box?? Do you need a path and texture name??

Has anyone done this before??

Any help would be very much apreciated.

Cheers

As I understand it the logic is the same.. if nothing but the camera is moving then you really don't need to rebuild the FG for everyframe.

I've never tested this myself though and I imagine that you only need to enter a patch if you are reusing a previously rendered FG data.

Final Gather is camera dependant, actually, so if you have an animation where only your camera is moving, Final Gather will still need to be rebuilt.

Turning off the rebuild FG option does not mean that FG will never rebuild. It is kind of misleading in that sense. What it does mean is that Mental Ray rebuilds whatever portion of the FG map that it needs to depending on the new circumstances of each frame, rather than rebuilding the entire FG map.

So, if you have an animation where the camera is NOT moving and there isn't a whole lot of movement within the scene, turning off FG rebuild should greatly speed up render times... however, this is a very, very rare circumstance.

When you do have camera moves or movements in your scene, with FG rebuild turned off MR refers to its saved FG map and then essentially fills in the holes in areas where the map does not match what it did on the previous frame...

So, will turning off FG rebuild speed up render times? It is very dependant on what is going on in your animation, but almost without fail, even with FG rebuild off you will still have a portion of your render times going to generating an FG map.

It might just be slightly faster than if you were rebuilding the entire map.

Its been my experience, though, that there isn't much difference, and I have actually had some odd quality issues stopping and then restarting renders using a saved FG map.

What I WOULD recommend, though, if you want to do a fly-through animation (only camera moves) is to set up some lights emitting photons, calculate once and turn off photon map rebuild so MR refers to its saved photon map and doesn't have to rebuild that, and then turn on "precompute photon look-up" in your FG settings. This saves some FG data in the photon map and will speed up FG. Usually this also slows down the photon phase, but since you are bipassing the photon phase by referring to a saved map, you get the benefits of the precompute photon map (faster FG) without the drawbacks (slower photons).

In theory, you could have tons of photons saved in a map that you only compute once, and then turn down the FG rays to a minimal and get a relatively fast render.

However, again, if you have things within your scene moving besides your camera, you probably want to rebuild the photon map.

Anyway, I hope that was helpful and not too long-winded ; )

-Steve

Thanks for the indepth explaintion!