glad to help ragingbull,
its imperative to talk about workflows when dealing with final gather because there are many ways to go about it depending on what you do. 3ds max workflows in their help files, tutorials, master classes, website demos, etc… are really aimed to arch viz. without moving objects.
i’ve seen every video in THE AREA and not many talk in depth about rendering motion graphics or production rendering techniques. gnomon workshop has awesome tutorials but also i think it needs more info in this matter. and by this i mean speed and good quality. I think jeff patton will do some new tutorials soon about final gather and moving objects.
for what i have read in many forums people have figured out rendering techniques on their own. final gather will definetly always make your renders slower no matter how optimized you get it to work.
From what I heard - max has an integrated fg-interpolation algorithm which creates several fgmaps according to the frames in the timeline and starts interpolating them - so flickering is reduced by a huge amount.
In maya - you don’t have that feature - I think someone made such a standalone interpolation application to use with maya. – but so far definitely every application is better than maya. but hey maya 2011 is out - maybe they optimized things at fg - help says a lot about gi optimizations but nothing about fg - but maybe they took the global term where gi means fg&gi - have not read it yet.
Sorath, you are correct 3ds max has that feature but as far as im concerned i cant get it to work nice. if you dont do a really high calculation you get flicker anyway. i guess im gonna look into it a bit more but for now i try to stay away from final gather as much as i can if the project doesn’t need it.
just saw saturns website in the lab section he has a great article that i think many need to read.
http://www.harrybardak.co.uk/diff_conv.htm
never saw that article before but thats the way i go about it in 3ds max tho a bit different
because max doesnt do shader connections like xsi or maya one needs to do it in a different way. i use the ao shader in an omni light(set to environment) as a lightshader that way i avoid having to composite it afterwards. you can have nice image based lighting that way too by putting some environment image in your background as spherical mapping (just like you do on any hdr) and in the AO shader settings enable environment Type=1. be sure to use the blur offset option to smooth the environment images otherwise you get awful noise. but no flicker! 
.
another underrated feature of final gather is that it can behave as ambien occlusion when used in brute force.
if anyone is wondering you achieve brute force sampling method by turning off interpolation (set it to 0) by doing that it no final gather no longer takes it into account so point density is not been used also. mostly now you are just adding samples to clean up the solution but it takes a hell of a lot of time to render, but thats because its sampling to infinity, just as an AO shader does when you dont limit the max distance parameter.
so if you limit the ray distance for FG you get an AO approach but with lightbounces and self illumination lighting, but no flicker :buttrock: .
for example i rendered this test to prove my point
HERE IS A VIDEO
video
is not physically accurate by any means but i was able to render it in really quick speed (about 2.2 minutes per frame in a slow 4 buckets machine) and still get all the benefits of FG but without flicker.
this is a worst case scenario for final gather and moving objects… dark interior with animated light. it has A&D materials, sss skin, hair and fur on the mittens and boots, animated self illumination on the wall, one sky portal light on the window, glass, etc… all of this would be a pain in the neck to composite thats why i think many could take advantage of this technique for complicated scenarios.
my settings were these
basically i turned to 0 all the options i didn’t need to illustrate that brute force doesn’t need a lot of tunning.
most important parameters are amount of rays and limit ray distance.
when using brute force the rays behave very similar to AO so the amount of samples you use for AO would do here too, tho you have to consider that this method calculate more than occlusion so for cleaner results more samples are needed but not that much actually.
then limit ray distance works exactly as AO max max distance. its how far fg calculates interaction between objects. and as low as you set it the faster it’ll be but less light interaction with objects you get. so this is one of the more important values to tune.
i used 75cm because it made the self illum light reach a nice distance. i didn’t have that self illumin then i might had set it lower.
bounces work the same as always so if you really need more light bounce increase the trace depth values accordingly.
one thing is important to note is that this method seems to make images brighter and thats because rays are not looking up into infinity so some objects are not shadowing areas that they should… just like AO behaves when you limit it.
Another important note. i didnt use image based lighting here, just flat colored skylight, when using images in environment samples need to be a bit higher. also really important to blur the images specially hdrs
here is an example with material override
could have use more rays but for a test i think its enough. also when objects have a lot of textures more grain is acceptable.
hope its useful for someone, it sure is helpful for me when i dont have time to do complex composites and have to render complex light scenarios.
