A basic compositing question


Hey guys,

I generally do a lot of motion graphics work but am getting more into vfx and just had 1 pretty basic question to ask, much appreciated if someone could please explain!

  1. I was watching the making of this tvc… http://www.cgsociety.org/news/article/3555/behind-the-scenes-100-cgi-bear-for-third-wheel-
    and am wondering what is the process of painting out the reference guy dressed in the bear suit? Obviously with a locked off shot you could shoot a background plate, but with a handheld shaky camera how is this done? Is it actually painted out frame by frame or is it more a matter of the 3D bear being slightly larger than the reference bear to sit over the top?

Thank you!


Hi Lachy,

It’s a nice piece! I’d imagine the suit was used for mocap data, or for scale and position reference if entirely hand animated - then yes the CG bear would be ‘laid’ on top, camera tracked into position, and roto’d behind other objects. I doubt it would have always covered the suit, so in places a bit of Mocha Pro or other 2D tracking was perhaps used to paint out the overlap.



Cleanplating is a whole discipline in compositing. It all depends on the complexity of the motion and decisions need to be made about reconstructing the background. Sometimes it’s as simple as tracking parts of the plate back over using mocha or point-tracking. It can get as complex as rebuilding the environment and reprojecting on 3D geo.
In the example you showed, there would certainly be some background cleanup, and probably quite a lot of paint work to clean up the actor in the places where there was interaction (the paws, etc.)
It’s a decision you’ll be making shot-per-shot.