Kirbosh wrote:It would be a video, Although if there is a constraint on moving angles of the head id like to know.
That constraint is set by your skill level. Anything is possible with enough time, you just don't want to stretch it that far.
There are simpler and dirtier solutions and hard fine solutions to this problem and many inbetweens. Simple dirty solution is to constrain the motion in footage to basic translation only and track a new faceless texture onto the face. Then blend it in on edges as necessary.
Hard proper solution is to do full 3D head object matchmove with deformations, create a new faceless head model and apply the original tracked motion to that rig, including deformations. Then shade, light and render the 3D head, with masks for blending regions etc and composite it on top of cleaned plate. For example you might need to get rid of some face features like nose etc if they extend outside the boundary of your new faceless face at some angles.
Full face is hard to cobble together quickly because the area that needs replacing is big and you can't clone the textures from surrounding areas as there aren't enough of suitable texture. Smaller subsets of face are for this reason easier, since you can get away with some smart cloning and blending of clone patches and strokes as long as the motion isn't too aggressive on perspective changes etc where keeping the appearance of some region becomes problematic due to big variation.
Mon Feb 12, 2024 9:53 pm