Re: Creating Depth (Where None Exists)

Yikes. Trees. Bad idea, but let's see what we can do.

Here's what I imagine:
layer 1 (top): the large foliage at left-to-center of screen. Use keying to pull out the dark tree trunks, masking to kill everything to the right of the right trunk.
layer 2: tree trunks. Fill in the tree trunks with a dark solid, or patches of tree trunk copied from elsewhere. Now here's the problem: with what are you going to fill the areas to the left of the tree trunks once layer 1 is removed? I would grab sky from another shot and dupe the grass under the talent and move it to the left. I wouldn't try to add any trees there unless it looked bare.
layer 3: talent: I hate diff mattes (they rarely work in real life) so I'd roto the talent -- but then again, maybe a diff matte would work around the body, but you'd probably have to roto his head and shoulders.
4. Background: Once you remove the tree trunks, how would you fill in the area behind? You've got to find trees from somewhere else because you can't create the edges of the existing BG trees by cloning. You have to create trees, with shape and all -- not just a tiled texture. You're making trees.

If I were given this shot, I'd convince the DP to do this if the talent (and bluescreen) was unavailable for a reshoot:
1. keep the talent and roto the BG out
2. shoot grass for the talent to stand on, with background trees but no foreground trees and bring him in over that layer
3. shoot various foreground trees with sky BG for keying out
... of course, the lighting would have to match. :-)
4. Comp the talent, BG and new FG trees in layers in AE.

Remember: these jobs are more than just removal, there's also filling in. Ask yourself how much needs to be filled in, and with what. How much needs to be created? The "kid stays in the picture" animated 2.5D stills effect requires limited cloning around edges due to reveals due to movement. This is different, no?

Anyway, hope that helps.

Tip: it sometimes helps to look at these scenarios with the right eye closed. This flattens things and gives a graphic perspective, which is all the camera sees. Keeping both eyes open and seeing in 3D might fool us into imagining that the layers actually exist in the camera when of course, they don't, and the job is harder than we think.