Also, to clarify, the issue is not with respect to the ftheta model.
I have tried my ftheta model with other h264 videos provided with Nvidia and it works “fine”, as in the rectification is not perfect but at least the rectified video makes sense.
When I am using my own h264 video, no matter which rig file/calibration I am using, the rectified video on the right side looks unreadable (basically junk).
There is no problem with the visualization of my h264 window on the left, its just the rectified window that has the problem.
I have tried converting the original yuv of my video in different ways to h264 but haven’t managed to get a different result yet.