Talk:Motion compensation

Merger Proposal
I propose that motion estimation be merged into motion compensation. An earlier section "Proposal: Separate Motion estimation" proposed, with justification, that they be separated, but now a page on opticla flow exists, in the presence of which motion compensation is redundant. Motion estimation can be redirected to optical flow instead. --Venkatarun95 (talk) 22:48, 16 June 2018 (UTC)

Context
I placed the context tag on this article having reached it via Random. The first line assumes far too much mentioning "frames" without any reference to video. Read the first sentence in isolation assuming you're not already discussing video:


 * Motion compensation is a way of describing the difference between consecutive frames in terms of the where each section of the former frame has moved to.

Now, as a general reader stumbling upon this, do I have a clue as to what it means? You've lost me and I move on to something else. It may be as simple as adding the phrase "...in digital video encoding..." after "Motion compensation" but it might be achieved other ways too. AU Tiger ʃ talk /work 18:10, 30 January 2006 (UTC)
 * I have added a new introductory paragraph, please let me know if this clarifies the context issue. This is my first wikipedia edit, not sure of all the protocols for making changes, and notifying people. TheRoan (talk) 04:23, 31 May 2008 (UTC)

means that the difference of the motion vector and the neighbouring motion vector(s) encoded before is encoded.

This line may need a little clarifying espcially the "encoded before is encoded" part.

Motion compensation
I found this sentence confusing:
 * As in the process of successively finding/refining motion vectors some neighbouring MVs are not known yet, the corresponding prediction errors can be ignored (not added) as a sub-optimal solution.

Tkho 05:57, 1 February 2007 (UTC)

Could be equal to the following (please verify the technical intent vs., the original).

"As in the process of successively finding/refining motion vectors, a given encoding time-budget for a frame may elapse. If the encoding engine needs to start encoding the next frame, some neighboring MV’s may not be known yet, so (to meet the time-budget), the corresponding prediction errors can be ignored. Not adding these errors will result in a frame being encoded quicker, however the video quality would be sub-optimal."

User: kmorris/unregistered, February 27, 2008 —Preceding unsigned comment added by 161.44.227.200 (talk) 21:52, 27 February 2008 (UTC)

polynomial complexity
"Therefore, there is no algorithm with polynomial computational complexity that guarantees optimal motion vectors" - maybe there is no linear algorithm, but it seems these algorithms are still polynomial in the input size.

Proposal: Separate Motion estimation
Motion estimation is a subject in it's own right. It seems strange to have it as a subsection of motion compensation. I propose to separate it to a new page and remove the redirect for Motion estimation. Any objections ? Kegon (talk) 03:25, 18 February 2008 (UTC)

Motion Estimation and Motion Compensation mixed up?
Is it possible that you often find those two terms used a bit mixed up? To me it seems that Motion Compensation is also used as a summary of both methods... Whereas there's actually a quite clear distinction between the two... HerrPi (talk) 08:00, 28 August 2009 (UTC)

rewrite illustrated example text
The example seems good but the accompanying text is hard to follow. Could someone who knows more about this subject than I rewrite it please? Da5nsy (talk) 18:42, 29 March 2020 (UTC)