Difference between revisions of "SoC 2009/Weighted P-frame Prediction"

From VideoLAN Wiki
Jump to navigation Jump to search
m
(Vadalism Undo revision 15317 by Talk))
 
(7 intermediate revisions by 3 users not shown)
Line 3: Line 3:
 
== Abstract ==
 
== Abstract ==
 
x264 is a highly popular h264 encoder. It currently does not implement the entire spec of h264. Currently weighted P-frames are not used to assist in prediction. If implemented this would give huge benefits in cases where the scene fades or where there are flashes. I am proposing to implement an implementation of this that is good enough and fast enough to warrant inclusion in most video encodings.
 
x264 is a highly popular h264 encoder. It currently does not implement the entire spec of h264. Currently weighted P-frames are not used to assist in prediction. If implemented this would give huge benefits in cases where the scene fades or where there are flashes. I am proposing to implement an implementation of this that is good enough and fast enough to warrant inclusion in most video encodings.
 +
 +
== Code ==
 +
My patches (and codebase) are [http://repo.or.cz/w/x264/x264-p-frames.git?a=shortlog;h=refs/heads/gsoc here] for now, but they may soon move to videolan's server
  
 
== Goals ==
 
== Goals ==
Line 8: Line 11:
 
! No. !! Task Description !! Status !! Comment
 
! No. !! Task Description !! Status !! Comment
 
|-
 
|-
|1 || Build weighting framework || style="background: #00ff00"|done || I have tested it to work.
+
|1 || Build weighting framework || style="background: #00ff00"|done || .
 +
|-
 +
|2 || Brute Force Optimal Weights||style="background: #00ff00"|done ||
 +
|-
 +
|3 || Try use these new weights for some interesting things to improve quality||style="background: #00ff00"|done ||
 +
|-
 +
|4 || Find Optimal Weights in a fast way (probably based on the findings in 3.) ||style="background: #00ff00"|done ||
 
|-
 
|-
|2 || Brute Force Optimal Weights||style="background: #ffff00"|In Progress || Basic algorithm is there, just not working optimally
+
|5 || DSP Functions ||style="background: #00ff00"|Done ||
 
|-
 
|-
|3 || Try use these new weights for some interesting things to improve quality||Not started ||  
+
|6 || Try duplicating reference frames ||style="background: #00ff00"|Done||  
 
|-
 
|-
|4 || Find Optimal Weights in a fast way (probably based on the findings in 3.) ||Not started || Actually I have got a lot of ideas, some of which I have tested. Just nothing using the current framework so it is possible that it is all useless.
+
|7 || Other weird uses ||style="background: #00ff00"|done || Implemented a k-means cluster algorithm to find a number (k) of good weights and use them
 
 
 
|-
 
|-
|5 || Optimize ||style="background: #ffff00"|Currently working on the weight assembler code. || Never done assembler before so this is going slooow.
+
|8 || Finish off small things ||style="background: #00aa00"|Working || Todo: Finish multipass, cleanup k-means a bit
 
|-
 
|-
|6 || See if explicit B-frame weighting can do good things ||Not Started||  
+
|9 || Optimize algorithms and test like crazy || Not started ||  
 
|-
 
|-
|7 || Other weird uses ||Not Started||  
+
|10 || Cleanup and code review || Not started ||  
 
|}
 
|}
 +
 +
===Current Issues===
  
 
== Timeline ==
 
== Timeline ==

Latest revision as of 06:23, 12 January 2010

This project is part of Google Summer of Code 2009.
Student: Dylan Yudaken
Mentor: Jason Garrett-Glaser

Abstract

x264 is a highly popular h264 encoder. It currently does not implement the entire spec of h264. Currently weighted P-frames are not used to assist in prediction. If implemented this would give huge benefits in cases where the scene fades or where there are flashes. I am proposing to implement an implementation of this that is good enough and fast enough to warrant inclusion in most video encodings.

Code

My patches (and codebase) are here for now, but they may soon move to videolan's server

Goals

No. Task Description Status Comment
1 Build weighting framework done .
2 Brute Force Optimal Weights done
3 Try use these new weights for some interesting things to improve quality done
4 Find Optimal Weights in a fast way (probably based on the findings in 3.) done
5 DSP Functions Done
6 Try duplicating reference frames Done
7 Other weird uses done Implemented a k-means cluster algorithm to find a number (k) of good weights and use them
8 Finish off small things Working Todo: Finish multipass, cleanup k-means a bit
9 Optimize algorithms and test like crazy Not started
10 Cleanup and code review Not started

Current Issues

Timeline

This is an estimation.

March 20th or so - April 3 Qualifying task. Was a first attempt at goal 1.

April 20 Students announce

April 21 - May Goal 1 redone better and acceptably.

May 1 - May 20 Work on goal 2

May 23 Program Starts

May 26 - June 12 Exams

June 13 - July 13 Code

July 13 Mid Term deadline. Goals for here have not been set but personally I would like to have goal 2 & 4 finished with goal 3 having a decent chunk of work done. This weird out of order thing is because goal 3 can be worked on continuously as it is not really set, more the trying of ideas.

July 13 - August 17 Code

August 17 Program end