Idea: if we only care about view-optimization, and we have our own metrics by simply measuring how much space the video takes up, we don't really need to involve the Oculus at all (it can be like theoretical research stuff), which we can then run asynchronously on latedays' Tesla K40 in CUDA
Idea: if we only care about view-optimization, and we have our own metrics by simply measuring how much space the video takes up, we don't really need to involve the Oculus at all (it can be like theoretical research stuff), which we can then run asynchronously on latedays' Tesla K40 in CUDA