I am looking at the detail of stereo rendering in Panda3D, and I would like to have a confirm about the actual stereo camera setup used.
In CG, usually, off-axis (asymmetric frustum) setup is the correct choice.
In Panda3D, stereo rendering is based on two parameters: interocular distance (no problem with that) and convergence. This can lead to think about a toe-in configuration; however, it is described as “distance from the camera to the area of interest in your scene”, where “area of interest” I presume it is “zero parallax”.
I looked also into the source code, and from what I see, there is no keystone correction (needed to correct vertical parallax introduced in the toe-in setup) applied in case of stereo rendering.
Moreover, the calculations lead me to think that Panda3D uses off-axis setup, using “convergence” as a parameter to calculate the “amount of asymmetry” in the two frustums.
I am not so confident with the source code of Panda3D, so I could have missed something.
Please, could anyone confirm (or correct) my conclusion?