@cnoellert - something useful like - here is a plane that represents the entire field of view, orthogonal to the focal plane of the recently tracked 3d camera.
attach your gmask to the axis of this plane and click and click and click and close. (sort of like the perspective grid)
too much math for a bear of very small brain, but not too much math for a bunch of Canadian bears…
That’s what I was thinking too… basically a clipping plane frustrum rig, representing the projection depth, which you could adjust would be super helpful. Just so things fall where intended.
@philm @cnoellert What about this kind of scenario: control select multiple points of the same plane, and like syntheye’s add card feature a gridded plane emerges that follows and extrapolates that perspective in space, and then attach gmask beneath that and points follow that plane in terms of depth at time of drawing. Wouldn’t be useful for everything, but definitely might be helpful for some tasks on more curvaceous camera moves.
Several months ago, I went down this rabbit hole doing roto on a tracked object with an animated camera. Having a locator on the object was a red herring. I got an excellent approximation of the item and was so pleased with the sense that with just a few tweaks, it would be perfect. But when it came time for ultra fine adjustments- The ONLY way I found to control the spline was using the vertices fields and the handles controls. It was like driving a racecar while being locked in the trunk watching a flipfone feed out the front.
I can now caution all that might see advantage in having a camera: maybe don’t. Gmasks in 2d will give you far better control and results in a reasonable amount of time.
Just an anecdote, but you do you.
Ha! No I’ve experienced this same pain many times! I generally will use my tracked camera to roto static objects in landscapes. But the issue is it won’t follow the rotation of the camera correctly because it’s drawn in the wrong depth. Thus having to painfully adjust the points as you’re describing above. Having a tool that at least lets you draw a mask relatively in the right depth I think would be helpful. The other option is obviously just drawing static masks and projecting them onto action geometry and tracking them on, but then any adjustments are upstream and painful. I don’t disagree that there are many instances where it’s not worth bringing the camera into the tracer!
One thing y’all could try instead of doing roto in 3D: 3D stabilize your plate on a surface placed at the coordinates of the object you want to roto. Feed that stabilized plate from Action into a Tracer and then that matte into an Action that projects it back onto that surface. You and up filtering the matte image so you’ll lose some detail, though