Current Status of HVC (High-Performance Video Coding) in MPEG
2009-07-03 H.265/HEVC View Comments Views(20,096)In the last MPEG meeting, MPEG issued a Call for Evidence (CfE) on High-performance Video Coding (HVC). Nine responses to the CfE are received in this meeting (89th MPEG London). Those reponse proposals adopt typical coding tools in KTA, such as adaptive loop filter (ALF), extended macroblock size (EMS), larger transform size (LTS), internal bit depth increasing (IBDI), adaptive quantization matrix selection (AQMS), as well as new tools, such as modified intra prediction, modified de-block filter, decoder-side motion vector derviation (DMVD).
The objective experimental results show that 20% average bit reduction is achieved compared with H.264/AVC High Profile for all classes of test video sequences (Class A: 19%, Class B:25%, Class C:22%, Class D: 15% bit reductions, respectively). Subjective evalution is also conducted during this meeting. The purpose of subjective evaluation is identifying examples that give the best evidence and assessing whether the evidence is large enough. The subjective evalution results also show positive evidence to support the availability of suitable technology to justify the start for a new standardization activity.
Therefore, MPEG plans to move forward toward quickly issuing a formal Call for Proposals (CfP)Â on HVC (as soon as October 2009). The timeline of CfP on HVC:
- July 20, 2009: Availability of test materials defined in Draft CfP
- August 15, 2009: Availability of preliminary anchors
- October 27, 2009: Deadline for pre-registration (mandatory)
- October 31, 2009: Final Call for Proposals (public)
- November 15, 2009: Formal registration
- December 15, 2009: Coded test material available at the test site
- January 1, 2010: Subjective assessment starts
- January 11, 2010: Registration of documents describing the proposals with MPEG video chair
- January 12, 2010: Submission of document to MPEG Video chair
- January 16, 2010: Report of the subjective test results
Permanent Link: Current Status of HVC (High-Performance Video Coding) in MPEG
# 2009-07-03 Friday 10:22 pm
Basically, no H.265 encoder until after 2010?
# 2009-07-04 Saturday 9:29 pm
Also, is the 20% bitrate reduction achieved by less computational complexity than H.264, the same, or more? This doesn’t satisfy the original goal of reaching 50%.
# 2009-07-06 Monday 1:43 pm
Of course, this 20% objective gain is achieved by more complexity than H.264/AVC. In addition, the goal of 50% bit reduction is for the same subjective quality, instead of objective quality.
# 2009-07-07 Tuesday 12:34 am
What’s the difference? And wasn’t H.265 supposed to gain 50% at the SAME complexity?
# 2009-08-05 Wednesday 1:51 pm
Is it means the battle between HVC and NGVC has been begun?
# 2009-08-06 Thursday 10:00 am
VCEG and MPEG have a disagreement on the way to collaborate, but I would not think it is a battle.
# 2009-08-07 Friday 9:26 am
I tested Kta2.4r1, but only got the about the 5% bitrate reduction for the same subjective quality. I don’t know whether my encoding parameters are wrong. Who could tell me the correct papameters or how to get them? Thank you very much!
# 2009-08-07 Friday 10:15 am
Did you follow the common test conditions specified by VCEG? The latest version is VCEG-AJ10. In the last meeting, performances of KTA were reported, as in VCEG-AL14 and VCEG-AL15.
# 2009-08-07 Friday 4:44 pm
I got no idea if any of you admins on this site are affiliated with MPEG/VCEG/whatever but I strongly urge all of you to carefully undergo research and not rush a release of a new standard. Take as long as you want, because whatever H.265 is supposed to be, it should be significantly more advanced than H.264. A 20% increase in efficiency is nothing, and only proves it to be an extension and not a new standard.
H.265 shouldn’t introduce concepts that linearly derive from its predecessor and only improve by a margin. It should be more INTELLIGENT than H.264, e.g. be able to compress cartoons with far greater efficiency than more complex ****. As an example, I had a recording of an old game with fast paced scenes accompanied with rain/snow. MPEG-2 even at 3.5 mbps displayed artifacts (the snow was choking the **** out of it) while H.264 was higher quality at only 350 kbps. While H.264 is only 3-4x better than MPEG-2 for real-life graphics, it was 10x better for this case. Understand what I mean?
H.265 may be the last video standard to be released as bandwidth will be way less of a problem once a nationwide fiber optic networks are built, and I do not wanna be stuck with a weak extension of a video standard that could’ve been much more. Don’t **** this up.
# 2009-08-07 Friday 8:42 pm
Me also I tested Kta2.4r1 with CrowdRun 1080p content as imput and I obtained only a about the 10% of bitrate reduction for the same objective quality. Is there somebody that can confirm this result?
# 2009-09-16 Wednesday 5:04 pm
Moreover, now with the advent of mb-tree in x264, there goes your 20% advantage. But this is an example of a more intelligent codec. It can discern one element in the video from another and the loss/efficiency is on a more variable level.
For something to supersede H.264, it would have to abandon the outdated DCT-based concepts and any kind of transform entirely. I’m not talking wavelet ****, as that itself is outdated and hasn’t proven to be any better than DCT transforms.
If “depth” in a 2D image can be somehow defined mathematically and a 3D codec is built then that would be your silver **** bullet in the video codec industry.