I am currently writing an AR experience. I have a Camera View Matrix, A Camera Perspective matrix and the Model Matrix of the content I display.
The AR content should can be switched to "2D", so it is attached to the screen.
For this, I have predefined Model, View and Projection Matrices.
My first idea was to calculate both MVPs and sending them both to the shaders and giving each shader an interpolation factor (I would interpolate the vertices)
This would work out I think, but I don't know if this causes deformations during interpolation.