[opendtv] Re: The next video codec is on its way...
- From: "Manfredi, Albert E" <albert.e.manfredi@xxxxxxxxxx>
- To: "opendtv@xxxxxxxxxxxxx" <opendtv@xxxxxxxxxxxxx>
- Date: Tue, 30 Nov 2010 17:32:15 -0600
Olivier Houot wrote:
> While some people are still arguing about the relative merits of
> MPEG2 and H.264 (no offense intended :-), the next video codec,
> HEVC, is being prepared.
>
> Once again, they are talking about a two-fold reduction on bitrate
> for the same level of quality, but now in comparison to H.264.
>
> The interesting thing is that a whole lot of objective and, more
> importantly, subjective testing has already been achieved on the
> new algorithms, and it appears to validate that claim.
>
> No revolution in sight for the underlying principles.
Good news. Now we can simmer down the hype about H.264.
Here's the way the new evolutionary update is described in September 2010 on
the Vcodex site:
http://www.vcodex.com/h265.html
"The initial Test Model has similarities to earlier standards such as
H.264/AVC, including block-based intra/inter prediction, block transform and
entropy coding. New features include increased prediction flexibility, more
sophisticated interpolation filters, a wider range of block sizes and new
entropy coding schemes. Coding performance varies across the different
proposals. It looks like we might expect to see a 2x compression improvement
compared with H.264/AVC (i.e. half the bitrate at the same visual quality), at
the expense of a significant increase in computational complexity (perhaps 3x
or more)."
As expected, the evolutionary upgrade (e.g. more block architectures) requires
higher power in processing. Doesn't say how that extra processing power
distributes to the encoding and the decoding process.
> This is all in the last issue of DVB-scene, together with
> several interesting articles about DVB-T2. They also mention
> tests up to 4096 QAM for DVB-C2...
http://www.dvb.org/news_events/dvbscene_magazine/DVB-SCENE36.pdf
The "objective" comparison shown in the DVB-T magazine, where the SNR of the
resulting image is compared with the original, shows that the codecs give more
similar results as the bit rate goes up. Which is what you'd always expect.
It's the "subjective" results that muddy up the picture, as it were. I don't
know why that happens, but I wouldn't be surprised if it's caused, at least in
part, by some artifacts which the subjects find appealing in the new codec.
Like for example, in audio codecs, enhanced high frequency content, even if it
wasn't actually meant to be there.
Bert
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
Other related posts:
