| |
 |
|
|
Movies Forum Index » Visual Effects Forum » Curious about old animation effect
Page 2 of 2 Goto page Previous 1, 2
|
| Author |
Message |
| Thad Beier |
 Posted: Mon Aug 20, 2007 12:06 pm |
|
|
|
Guest
|
D&T wrote:
Quote: I'm doing some amature wire removal and am having probs with getting the
background to stay still. I've tried these methods already:
*.flm to Photoshop and repaint. (not a consistent matte)
AE and copy a bit of "good" background and copy over "bad" background.
and Combustion 3 doesn't work with ME so I am waiting for an XP purchase.
SO... how do the professionals do it with out the inhouse software?
Duncan
There are many techniques, but I don't know if any of them are free.
We do a lot of rotoscoping, either in Shake or in our in-house roto tool. Basically, you draw an outline with some kind of spline curve, then go to another frame, and edit the curve to fit, and keep doing that until you are satisfied with the results or you kill yourself. We have recently been using a lot of tracking-assisted roto, where we first track the shot in 2D or 3D, and use that to move the curve into more-or-less the right place at a frame, it makes the hand-editing easier. Not easy, mind you, but easier.
Once you have the area of interest identified with the curves, the programs generate a matte, which can be used by a compositing tool to layer the fg over your new bg.
I have to say that this is fairly stone-age technology. There is some 21st century technology available, that was have yet to use. There's a program called KnockOut (and now KnockOut2) that does an intelligent separation of the fg from the bg, based on a fairly loose hand-rotoscoped curve. This program, and others like it, do statistical analysis of the clearly-bg and clearly-fg parts of the image (identified by the user by drawing this loose roto) and then on a pixel-by-pixel basis calculating the percentage of coverage of the fg. There were some results published at the recent Siggraph from the University of California at Berkeley, University of Washington, and a nearby large company that took that technique to a truly unbelievable new level, they called their research framework "Soft Scissors". You can look it up at
research.microsoft.com/~cohen/SoftScissorsFinal.pdf
I believe that the future of rotoscoping lies with these techniques.
Thad Beier |
|
|
| Back to top |
|
| |
Page 2 of 2 Goto page Previous 1, 2
All times are GMT - 5 Hours
The time now is Sun Jul 20, 2008 4:09 am
|
|
