backtop


Print 70 comment(s) - last by augiem.. on Aug 17 at 5:55 PM

New compression standard could be in commercial products as early as next year

The Moving Picture Experts Group (MPEG) met recently to issue a draft international standard of a new video compression format offering twice the performance of current standards. The new video compression format is called High Efficiency Video Coating or HEVC. The new H.265 compression codec is roughly twice as effective as the current H.264/AVC standard.
 
“There’s a lot of industry interest in this because it means you can halve the bit rate and still achieve the same visual quality, or double the number of television channels with the same bandwidth, which will have an enormous impact on the industry,” says Per Fröjdh, Manager for Visual Technology at Ericsson Research, Group Function Technology, who organized the event as Chairman of the Swedish MPEG delegation.
 
H.265 could usher in ultra high definition television with significantly more clarity than the 1080p we have today. The new compression format will also significantly reduce the bandwidth required for streaming video on mobile networks where wireless spectrum is at a premium. The format will pave the way for wireless carriers to offer more video services within the confines of their available spectrum.
 
“Video accounts for the vast majority of all data sent over networks, and that proportion is increasing: by 2015, it is predicted to account for 90 percent of all network traffic,” Fröjdh says.
 
He believes that the HEVC format discussed during the meeting in Stockholm could find its way into commercial products as early as 2013.
 
“It will take time before it’s launched for a TV service, but adoption is much quicker in the mobile area, and we’ll probably see the first services for mobile use cases next year,” Fröjdh added.

Source: Ericsson



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Encoding times?
By aegisofrime on 8/16/2012 11:21:16 AM , Rating: 2
As someone who uses x264 on a daily basis, what I want to know is how this halving of bitrate affects encoding times. There's probably no point if encoding times were to double.




RE: Encoding times?
By int_21h on 8/16/2012 11:27:03 AM , Rating: 2
No doubt. This will be very valuable for things like netflix and other streaming functions, but the encoding/decoding may be prohibitive.


RE: Encoding times?
By titanmiller on 8/16/2012 2:05:02 PM , Rating: 2
Computing is dirt cheap now a-days.


RE: Encoding times?
By FaaR on 8/16/2012 11:32:19 AM , Rating: 3
You wouldn't use a codec that offers same visual accuracy at half the bitrate because encoding (something you do only once) would take longer, even though playback - which is repeated many many times - would benefit greatly?

That's ludicrous.

Besides, hardware encoding acceleration is becoming more and more common, and work will of course continue on this front as well as time progresses.


RE: Encoding times?
By SlyNine on 8/16/2012 11:39:12 AM , Rating: 2
Still haven't seen a good X264 encoder.


RE: Encoding times?
By SlyNine on 8/16/2012 11:39:33 AM , Rating: 2
hardware encoder that is.


RE: Encoding times?
By XZerg on 8/16/2012 2:29:13 PM , Rating: 2
that depends on how resource intensive the .265 is when decoding. if very intensive now then eventually hardware will be fast enough but that's what will slow down the adoption even when the compression efficiency is so great.


RE: Encoding times?
By augiem on 8/16/2012 2:48:33 PM , Rating: 2
Well it just makes sense, doesn't it? When did you ever see a new standard in technology come out where the whole playing field was ready to go running with it? Standards are always developed with a mind toward the future, as the next step of evolution. Then when that standard is no longer sufficient for the growing needs, a new one is developed that has some room to grow into.


RE: Encoding times?
By Sivar on 8/16/2012 12:53:13 PM , Rating: 2
You may find this interesting:

http://x264dev.multimedia.cx/archives/360

"After a full 6 hours, 8 frames had encoded. Yes, at this rate, it would take a full two weeks to encode 10 seconds of HD video."
Note this was talking about an experimental, non-optimized, pre-h.265 encoder, but it gives an idea of the degree of possible performance changes.

Reducing media encoding size at a given quality is usually an exponential problem, as you have probably discovered using "placebo" mode for h.264, that's pretty bad.


RE: Encoding times?
By TakinYourPoints on 8/16/2012 1:28:37 PM , Rating: 2
You know what else took obscene amounts of time to encode for a little while? mp3s.

The first encoders I used in the 90s were slowwwwwwww, ridiculously slow, but they got faster. Encoder performance will increase, give it time.


RE: Encoding times?
By Sivar on 8/16/2012 3:41:39 PM , Rating: 2
True, encoders will get better and faster, just as they did with h.264.
I'm willing to bet that those first, initial encoders would run quite fast on a modern processor, though. In fact, the Xing codec of the day was faster than LAME is now. Of course, it produced offensively awful sound quality.


RE: Encoding times?
By bupkus on 8/16/2012 2:18:47 PM , Rating: 2
quote:
"After a full 6 hours, 8 frames had encoded. Yes, at this rate, it would take a full two weeks to encode 10 seconds of HD video."

I'd be willing to contribute cpu cycles a la protein folding.


RE: Encoding times?
By Guspaz on 8/16/2012 5:16:03 PM , Rating: 2
Holy out of context quote, Batman!

He was testing one of the proposal submissions (from Samsung/BBC) for h.265, back when various organizations were submitting their proposals for what the standard should include. Not only was this not the actual h.265 spec, but the encoders for the submissions were all tuned to spend ludicrously large amounts of time doing exhaustive analysis to provide optimal image quality (and score higher in "competition").

The actual h.265 draft is a mixture of aspects from various submissions, and while it will be significantly more processor-intensive to encode than h.264, it won't be anywhere near that bad.

This is like judging the performance of x264 based on the "placebo" preset (which does exhaustive motion searches), when nobody actually uses that in practice and it's something like ten times slower. Except, instead of ten times slower, try like 10,000x slower, since they tuned every aspect of their submission for quality over performance.


RE: Encoding times?
By Sivar on 8/16/2012 6:21:07 PM , Rating: 2
You are totally correct, and I did mention this:

quote:
Note this was talking about an experimental, non-optimized, pre-h.265 encoder


As an aside I do, in fact, use x264 placebo mode and have encoded over 1,000 hours of video doing so. I only get 1 - 3% smaller file sizes, but weeks usually pass between buying and watching a video, so I prefer the 30GB or so of total disk space savings.

Also of minor technical interest is that placebo mode it doesn't actually try every possible predictor combination ("exhaustive"); it does a Hadamard–Rademacher–Walsh comparison (sum of differences between each motion vector) which has the same result but is much faster (though still slower than a dead slug stuck in frozen molasses).


RE: Encoding times?
By Jeffk464 on 8/16/2012 3:50:23 PM , Rating: 2
It also means you need more cpu horsepower to decode/play the video. My E350 system can't handle hulu or netflix in HD, but does just fine on 1080p broadcast or blu-ray.


RE: Encoding times?
By Guspaz on 8/16/2012 5:23:15 PM , Rating: 2
That is a factor of your hardware acceleration being broken, not CPU horsepower. There are known issues with Brazos (the E350 is a Zacate chip in the Brazos family) and Silverlight, which Netflix uses, even though Brazos supports VC-1 acceleration (the codec Silverlight uses).

Blu-rays, which are typically encoded with h.264 and sometimes VC-1, would work fine with hardware acceleration.

When using commercial video playback software on a modern computer, it's very rare that you're actually decoding anything in software. Heck, even with MPC-HC, it'll default to DXVA hardware acceleration if supported.


"A politician stumbles over himself... Then they pick it out. They edit it. He runs the clip, and then he makes a funny face, and the whole audience has a Pavlovian response." -- Joe Scarborough on John Stewart over Jim Cramer











botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki