When trying to perform post-process shearing, wouldn't you need to know the rate at which the user's head is moving to properly create an inverse shear? Does this mean that based on the initial acceleration of the head, the image is immediately sheared to compensate? Or are these values precomputed and a linear interpolation is done between preocomputed shear amounts relative to head motion?

Same question as ^ -- predicting head motion sounds pretty unreliable to me. I never tried a VR headset before but I was curious what it looks like in the headset when you rapidly shake your head