Tweet
Hi,
I have a couple of doubts regarding the Motion Vector Prediction in H264.
1) As per the algo, i the top-right neighbour is not present, we consider the Top-left neighbour. Why is the Top-right given priority over Top-left in the first place? Is there any specific reason for that
2) Also, can the Motion Estimation be a lossy process if the MV prediction is incorrect. Say, my MVP should have been +10 and it has been taken as +8 and the difference encoded and sent. What I can understand is if the MVP calculation at decode side is done and the MVP comes to be +10, then the decoded picture might not be the correct representation. Can there be any other cases?
Thanks in advance
I have a couple of doubts regarding the Motion Vector Prediction in H264.
1) As per the algo, i the top-right neighbour is not present, we consider the Top-left neighbour. Why is the Top-right given priority over Top-left in the first place? Is there any specific reason for that
2) Also, can the Motion Estimation be a lossy process if the MV prediction is incorrect. Say, my MVP should have been +10 and it has been taken as +8 and the difference encoded and sent. What I can understand is if the MVP calculation at decode side is done and the MVP comes to be +10, then the decoded picture might not be the correct representation. Can there be any other cases?
Thanks in advance