Apple feedback for ImagePlane support

[dgovil]

Hey folks,
I'm just using this issue to collect our feedback for the ImagePlane proposal.

In no particular order:

1. It would be useful if the image plane was allowed to include its own depth information. This allows us to not specify the depth of the image from the camera as a single constant, but have it be pre-defined and vary across the frame. This is a workflow I developed at Imageworks and presented as Animation collaboration with depth compositing
   
   However, this then brings up the question of normalization of depth in images. Personally, I prefer unnormalized depth, but perhaps there could be attributes to help remap normalized depth?
2. The Positioning/placement strategy could perhaps be clarified. It states it's relative to the film back, but it's unclear if this uses the image corner or center, and if the values are normalized or not. Would changing image or aperture aspect ratios cause placements to change for example?
3. How might this work with multi-view images? Is the primary view to use implicitly derived from the camera role, or could there be an option to explicitly name the primary layer/aov/view to use?

[dgovil]

Some more thoughts:

1. How would image sequences be handled? I think image sequences are perhaps a somewhat underspecified area of USD, so this might be a good opportunity to explicitly describe their handling.
2. Is it possible to have more than a single image plane per camera? Often in VFX we'd want to place plates at different depths, such as a foreground and background plate.
3. Is there a method to enable/disable the image plane for rendering? Would that be Hydra side? Should there be something on the render output configuration to say whether you want the ImagePlane to be included in the output or not? e.g playblasts, an animator probably wants the image plane. final output, you probably don't want it there.

All of this perhaps leads me to believe that it could be its own prim that the camera has a rel to?