Quote from: Oshyan on January 19, 2013, 04:44:28 AM
Or you could just composite the output from TG and the output from 3DS Max/Maya/whatever, like the pros do. They seem to get pretty good results. Either way has its challenges, but the compositing workflow is probably more well documented, and better facilitated by the recently-added FBX support.
- Oshyan
On this I cant agree with you Oshyan. For Hannes' bird test, yes you are right. But imagine fireflies or butterflies in a prairie, with camera parallel to the tips of the wiled grasses. The creatures flutter in float up and down, in and out... The amount of masking and rotoscoping alone would make me puke. Add to that everything else, and forget it.
Even just take the OP image that started all of this. The usual way means that Hannes has to mask all of that foreground stuff out, and then set up multiple blend layers over edges. Or he would have to do a bunch of different passes in each software. Do we have z-pass yet?
Without development in this area, TG2 is background or foreground (in part) but not both at the same time, unless its pure landscape.
Remember chris_x422's post on his work in snowWhite. THat works just like you say, but only because nothing in the foreground is obstructing the view of the background. And I think even big productions don't like spending time on masking and roto, if they don't have to.
This is a bridge! If it can be made to work well up close in complex ways, it will mean one render sequence, not two. And a ton of other work cut out also. See? Or am I forgetting something important?