I have created several new shapes using OpticalFlowWrapping and exported the .png texture files for each through TransferTexture. When I take these texture files into Photoshop and switch between them, I notice the black marker placement on the face/texture is not 100% accurate between the .png files. Is there a way to better stabilize these black marker points between texture files?
Yes, OpticalFlow is not 100% automatic solution and for complex cases requires setting control points. I would suggest selecting control points for markers.
This summer we are planning to release a set of tools that may significantly simplify this step by using machine learning for eyelids and lips detection.
It’s the information I’m looking for as well.