H.264 dead?
 

No sooner have transmissions in H.264 began that H.264 looks like it is soon
going to be replaced by a derivative of MJPEG2000 .mj2 compression, the new
standard in professional video and movie production, but which unlike
MJPEG2000 compares differences in motion between frames.

JPEG2000 already offers a 20% improvement over JPEG which is the root
compression system of H.264 but unlike the nasty blocky artefacts left by
over compressed or even moderately compressed JPEG images, the artefacts
created by JPEG2000 are closer to analogue blurring in optical systems,
which are a hell more desirable than the block and phantom colour
distortions of JPEG and H.264.

Click on these images for comparison.

http://en.wikipedia.org/wiki/File:JP...Comparison.png

http://en.wikipedia.org/wiki/File:JP...onstration.png

Now whereas the blocky artefacts produced by JPEG based video compression
are totally intolerable in shows like the Eurovision Songs Contest, ringing
artefacts caused by JPEG2000 would barley be noticeable on the Eurovision
Songs Contest because they would be masked by the motion blur of the
performers which would have caused huge blocking artefacts on JPEG based
compression.

So what are the chances of an open source project creating a motion
estimation enabled JPEG2000 wavelet compression based video codec before the
digital switchover in 2012 thus rending everyone's new HD equipment
obsolete?

H.264 dead?
 

Agamemnon wrote:

Snip

So what are the chances of an open source project creating a motion
estimation enabled JPEG2000 wavelet compression based video codec before
the digital switchover in 2012 thus rending everyone's new HD equipment
obsolete?

http://www.bbc.co.uk/rd/projects/dirac/index.shtml

H.264 dead?
 

On Sun, 1 Mar 2009 01:53:50 -0000, "Agamemnon"
wrote:

No sooner have transmissions in H.264 began that H.264 looks like it is soon
going to be replaced by a derivative of MJPEG2000 .mj2 compression, the new
standard in professional video and movie production, but which unlike
MJPEG2000 compares differences in motion between frames.

JPEG2000 already offers a 20% improvement over JPEG which is the root
compression system of H.264 but unlike the nasty blocky artefacts left by
over compressed or even moderately compressed JPEG images, the artefacts
created by JPEG2000 are closer to analogue blurring in optical systems,
which are a hell more desirable than the block and phantom colour
distortions of JPEG and H.264.

can you provide sources for this?

all the JPEG 2000 i have seen in use seems to be for contribution, and
for standard def TV runs at 10 Mbps+, and HD needs a fair bit more.
Some broadcasters are using 25 Mbps to get high quality.

The EBU was pushing it for contribution because it withstands lots of
repeated code / decode cycles well - but it isnt anywhere near as
bandwidth efficient as MPEG4 / H.264

Click on these images for comparison.

http://en.wikipedia.org/wiki/File:JP...Comparison.png

http://en.wikipedia.org/wiki/File:JP...onstration.png

Now whereas the blocky artefacts produced by JPEG based video compression
are totally intolerable in shows like the Eurovision Songs Contest, ringing

MPEG?

artefacts caused by JPEG2000 would barley be noticeable on the Eurovision
Songs Contest because they would be masked by the motion blur of the
performers which would have caused huge blocking artefacts on JPEG based
compression.

So what are the chances of an open source project creating a motion
estimation enabled JPEG2000 wavelet compression based video codec before the
digital switchover in 2012 thus rending everyone's new HD equipment
obsolete?

Dirac is another wavelet based scheme, so shares the property of
images blur as higher higher compression is applied.

As another poster says Dirac was developed by BBC Research, and is
offered as "no cost of licencing" to manufacturers.
http://www.bbc.co.uk/rd/projects/dirac/licensing.shtml

Note that might not be the same as "no licencing involved"....
--
Regards

- replace xyz with ntl

H.264 dead?
 

"Peter Watson" wrote in message
...
Agamemnon wrote:

Snip

So what are the chances of an open source project creating a motion
estimation enabled JPEG2000 wavelet compression based video codec before
the digital switchover in 2012 thus rending everyone's new HD equipment
obsolete?

http://www.bbc.co.uk/rd/projects/dirac/index.shtml

Brilliant. That's H.264 dead then, no royalties to pay for a start, and the
creators seem to prefer OGG over AAC too so that's AAC dead as well. The
only problem is that all the source code seems to be for Linux and there's
no ACM encoder that VirtualDub can use on Windows.

H.264 dead?
 

"Stephen" wrote in message
...
On Sun, 1 Mar 2009 01:53:50 -0000, "Agamemnon"
wrote:

No sooner have transmissions in H.264 began that H.264 looks like it is
soon
going to be replaced by a derivative of MJPEG2000 .mj2 compression, the
new
standard in professional video and movie production, but which unlike
MJPEG2000 compares differences in motion between frames.

JPEG2000 already offers a 20% improvement over JPEG which is the root
compression system of H.264 but unlike the nasty blocky artefacts left by
over compressed or even moderately compressed JPEG images, the artefacts
created by JPEG2000 are closer to analogue blurring in optical systems,
which are a hell more desirable than the block and phantom colour
distortions of JPEG and H.264.

can you provide sources for this?

http://en.wikipedia.org/wiki/File:JP...Comparison.png

http://en.wikipedia.org/wiki/File:JP...onstration.png


all the JPEG 2000 i have seen in use seems to be for contribution, and
for standard def TV runs at 10 Mbps+, and HD needs a fair bit more.

That depends on the quality that you want.

The people marketing H.264 are claiming that it is 5 times more efficient
than M-JPEG and twice as efficient as MPEG-4/Xvid/DivX. DivX are claming
that DivX is 60% better at compression than MPEG-2, which means MPEG-2 is
only 1.6 times more efficient than JPEG. Since JPEG2000 is 1.2 times more
effiecant than JPEG there is practically no difference between JPEG2000 and
MPEG-2, so why both with MPEG-2 which becomes totally useless when there
lots of motion, camera zooming and panning and flashing lights.

In my tests the H.264 implementation in ffdshow is worse at compressing live
musical performances than Xvid at the same bitrates and this is using the
MPEG profile for Xvid not the H.263 profile. So forget the technical and
marketing claptrap claiming H.264 is 2 times as good as Xvid or DivX.

H.264s main advantage over its predecessors is its Average Bitrate
implementation. It hasn't even implemented JPEG compressing as well as Paint
Shop Pro so what uses is all the motion estimation and prediction when the
underlying DCT compression algorithm hasn't even exhausted the limits of
JPEG.

Even at 2Kbps H.264 is no better or actually worse than the original MPEG-2
encoder that produced the content for ITV4 at 2Kbps unless the content is
mostly static, and forget about trying to use low bitrates to archive your
VHS collection. VHS contains so much noise and instability that you have to
use very high bit rates if you want to preserve the quality and not
introduce artefacts.

H.264 is basically a codec for internet streaming. That's what it was
designed for. You have to go below 3Kbps to see any advantage for it over
MPEG-2 or use it for low quality HD channels.

If you want high quality you might as well use MJPEG2000. A well implemented
MJPEG2000 codec should be able to handle 10Kbps better than either H.264,
H.263 or MPEG-2 can handle 8Kbps and what's the difference between these bit
rates, almost nothing. Once you start adding motion estimation then a
wavelet based system will kick the pants off any of these codecs even at
1Kbps on SD compared to 4Kbpos for MPEG-2 or H.264.

Some broadcasters are using 25 Mbps to get high quality.

If you are recording stuff like pop concerts 4Kbps is not even adequate for
SD. You need at least 8Kbps for MPEG-2 and using H.264 will not make much
difference in quality if its a pop concert, so you would really need 50Mbps
for HD.

With a wavelet based system blocking artefacts will be replaced by blurring
which means you could broadcast SD at 1Kbps and HD at 3Kbps, less than
current SD transmission, and not notice any difference between that and a
12Mbps H.264 HD stream since square boundaries would be replaced by blurring
and the blurring would be masked by motion blur, which would not mask square
boundaries.

The EBU was pushing it for contribution because it withstands lots of
repeated code / decode cycles well - but it isnt anywhere near as
bandwidth efficient as MPEG4 / H.264

When you are making movies and TV shows you want perfect picture quality
even when the image is moving. 1TB external HDDs can now be had for £60
which is substantially less than the cost of a Professional HDCAM or D1
video tape, so what it the point of using H.264? You might as well use
Huffyuv in YUY2 mode which will give you a compression factor of about 3 and
which is almost lossless and will give you 30 hours of SD on a 1TB drive,
whereas a HDCAM tape will only give you 40 or 50 minutes.


Click on these images for comparison.

http://en.wikipedia.org/wiki/File:JP...Comparison.png

http://en.wikipedia.org/wiki/File:JP...onstration.png

Now whereas the blocky artefacts produced by JPEG based video compression
are totally intolerable in shows like the Eurovision Songs Contest,
ringing

MPEG?

Yes. What's worse on satellite the bit rates are only 3,000 kbps and the
picture resolution is only 544x576 which makes everything look hideous.
Remember the Children's Eurovison on ITV a few years back which was
completely dire during the performances, and I'm not just talking about the
songs.


artefacts caused by JPEG2000 would barley be noticeable on the Eurovision
Songs Contest because they would be masked by the motion blur of the
performers which would have caused huge blocking artefacts on JPEG based
compression.

So what are the chances of an open source project creating a motion
estimation enabled JPEG2000 wavelet compression based video codec before
the
digital switchover in 2012 thus rending everyone's new HD equipment
obsolete?

Dirac is another wavelet based scheme, so shares the property of
images blur as higher higher compression is applied.

Which is completely tolerable whereas blocking is intolerable and
unbearable. The only time it distortion would kick in on normal video
transmission bit rates is in content such as the Eurovision Song Contest or
Athletics is when there is lots of action and flashing lights and those
mosaic video screens behind the performers producing a moving chequerboard
pattern as the camera pans across them. Wavelet based compression would only
cause blurring which would occur anyway on motion going at that speed so
would look natural, whereas blocking from DCT compression does not look
natural and unlike blurring has sold edges and is therefore easily picked up
by the human eye.


As another poster says Dirac was developed by BBC Research, and is
offered as "no cost of licencing" to manufacturers.
http://www.bbc.co.uk/rd/projects/dirac/licensing.shtml

Note that might not be the same as "no licencing involved"....

The project was handed over to the open source community.

--
Regards

- replace xyz with ntl

H.264 dead?
 

In article
wrote:
can you provide sources for this?

Stephen, look who you're replying to. Don't expect rational on-topic
argument.

--
I'm using an evaluation license of nemo since 62 days.
You should really try it!
http://www.malcom-mac.com/nemo

H.264 dead?
 

Agamemnon wrote:
[Snippety snip]

H.264 is a required blu-ray codec (as is AAC), and so neither are going
anywhere.

H.264 dead?
 

"David Glover" wrote in message
...
Agamemnon wrote:
[Snippety snip]

H.264 is a required blu-ray codec (as is AAC), and so neither are going
anywhere.

Blu-ray is already dead. Why would anyone buy a Blu-ray player for their
computer when they can get a 1TB drive for £60 now which will store over 30
times as much information as a Blu-ray disc which costs £20 and is faster to
access. In a few months the cost of 32GB data sticks will fall to less than
a Blu-ray disc so Blu-ray will be well and truly buried. Soon people will be
downloading almost all HD movies over the internet. With wavelet based
compression you should be able to fit a 2 hour HD movie into 6 GB.

H.264 dead?
 

"Agamemnon" wrote in message
. uk...

"David Glover" wrote in message
...
Agamemnon wrote:
[Snippety snip]

H.264 is a required blu-ray codec (as is AAC), and so neither are going
anywhere.

Blu-ray is already dead. Why would anyone buy a Blu-ray player for their
computer when they can get a 1TB drive for £60 now which will store over
30 times as much information as a Blu-ray disc which costs £20 and is
faster to access. In a few months the cost of 32GB data sticks will fall
to less than a Blu-ray disc so Blu-ray will be well and truly buried. Soon
people will be downloading almost all HD movies over the internet. With
wavelet based compression you should be able to fit a 2 hour HD movie into
6 GB.



I increasingly tend to use SD cards, especially as my local Novatech now
sell 4GB ones for around £8.. As very compact DVD players with SD card
slots, USB connectors and HDMI outputs can be purchased for around £30 I've
managed to persuaded several friends and family to invest in one (I have
also given a couple as presents) as it makes a reasonable substitute for the
now almost defunct VHS when exchanging programme material, the fly in the
ointment of course being the inconvenience of transferring data, which
usually requires the use a computer.
I can't really see any technical reason why PVRs don't have the ability to
be able to transfer program material onto SD cards using an agreed standard
such as mp4 or divx, something which probably could be easily incorporated
in the main menu software for a few pounds extra.

H.264 dead?
 

On Mon, 02 Mar 2009 18:35:56 +0000, Agamemnon wrote:

Why would anyone buy a Blu-ray player for their computer when they can get a
1TB drive for £60 now which will store over 30 times as much information as
a Blu-ray disc which costs £20 and is faster to access.

Two reasons --

1) To play Blu-Ray discs purchased or rented at the supermarket / video store.
Many people are still on 2 MB/s broadband or less and have no desire to pay
for a faster speed, even if there telephone line will support a higher speed,
or for connecting to cable TV Internet if available.

2) Portability of data to take around to a friend or send in the snail mail.

And it should be remembered that data sticks are not reliable long term storage
devices.