[Jpeg-users] Fwd: 20 Years of JPEG Celebrated With Software Launch

[mathieu <mat...@gm...>]

Just FYIing to myself...


---------- Forwarded message ----------
From: Guido Vollbeding <gu...@jp...>
Date: Jun 8 2006, 11:15 am
Subject: 20 Years of JPEG Celebrated With Software Launch
To: comp.compression


Thomas Richter wrote:

> That means, I still cannot use Q15 for proprietry products

You can use Q15 as part of that ITU Recommendation for any kind of
products
you want!  That is noted onhttp://ftp3.itu.ch/jpeg1x/.

> Though it would still not be JPEG, it's a different standard - to be
> precise an ITU-recommendation - based on JPEG resp. the identical
> ITU-recommendation. That doesn't mean much, of course, it's just a
> different number and a different organization.

It is the *real* JPEG organization, based on the *real* JPEG standard!
It is the *ISO* part of JPEG which went wrong and got corrupted and
created
a totally different animal "JPEG-2000" which has nothing to do with
the
original JPEG standard!
J2K is a totally different technique, they just were clever enough to
apply the name "JPEG" to it because JPEG was so successful.  They
could do
this because at that time (in the mid and late ninetees) the original
JPEG
inventors had largely left the JPEG committee to do other things, and
there
was a "vacuum" to fill.  So basically JPEG-2000 is a hoax.

> Beating JPEG is actually easy. The standard is a bit old, and an overall
> renovation would be suitable.

Wrong!  There were several attempts in the past to beat JPEG, and they
all
have failed, and they *must* fail, because none of the challengers
have
understood the basic JPEG (DCT) fundamentals.

> Back in the old days, a couple of
> compromises had to be made to make it suitable for the hardware back
> then. The question really is, why drilling up an old working horse that
> does the job fine by introducing marginal improvements that generate
> bitstreams that are not backwards-compatible if instead we can throw it
> away and start from scratch.

Thomas, sorry, but here you show your complete ignorance!
The opposite of what you say is true:  The original JPEG DCT approach
is
the *only* appropriate solution for image coding!  Of course you can't
recognize this because you have not understood or ignore the basic DCT
fundamentals.
You will never achieve any advance without basic understanding of DCT
fundamentals, and that's why all those attempts must fail.
And of course WE WILL be backwards-compatible:  our new decoders will
seamlessly read "old" JPEG files without any problem, the encoders can
optionally output old streams, or you can transcode between different
flavors.
Backwards-compatibility is exactly what YOU broke with your JPEG-2000
thing!

> I haven't compared the Q15 implementation with the QM implementation, so
> I cannot present numbers right now. Given that both are approximately
> comparable, the mentioned improvement isn't able to match with even
> existing standards today, as measurements on QM show. So the question
> really is, why do that? However, I'm with the ITU in the sense that I
> see no point in *not* defining a standard, so let's see how it goes.

Thomas, it is said in the message:  This is only the *first* phase
result,
and "Experts from SG 16 say to stay tuned for further developments."!

These "further developments" will be more important than that first
result!
You can find something about this in my contributed proposal, latest
Revision 3:

 http://jpegclub.org/temp/

  direct:http://jpegclub.org/temp/ITU-T-JPEG-Plus-Proposal_R3.doc

But of course if you read this, and if you would understand it, then
you
could not continue to propagate your misleading statements.
So better keep on ignoring it if you want to maintain your opinion.

> I also haven't compared with the new Microsoft code yet, I just don't
> have it.

I have it!
But I can't tell about the details here because I had to sign kind of
non-disclosure agreements.

> A new round of comparisons would really be suitable and I would
> be curious to see the results with Microsoft's code. One thing for sure,
> the 50% higher compression than JPEG as advertised by M$ is definitely
> marketing hype. I would belive it's better than JPEG-1 (no blocking),
> maybe it's better than JPEG2000, it would require testing to make
> statements. Statements like "They cannot!" *at this time* come
> definitely too early and are as much marketing hype as the hot air from
> Redmond.

JPEG-2000 is obsolete and will lose even more ground (if it ever had
any)
after this WM Photo attempt, because WMP is more related to
traditional
JPEG technology than JPEG-2000/Wavelets.
The developers of this WM Photo are, similar to JPEG-2000 proponents
and
other folks, not aware of the basic DCT properties for image coding.
Therefore they *cannot* beat our actual JPEG-Plus approach!
The original JPEG-1 was only an initial approach.  They found the
right
technique (DCT), rather by accident so to say, but they had no basic
understanding of DCT fundamentals, and thus couldn't exploit all of
its
potential.
That is exactly what we do now with our JPEG-Plus approach:
For the first time present an appropriate exploitation of basic DCT
properties and thus approach its full potential!
And I can assure you that our results are superior to anything what
you have seen so far.  As said:  Stay tuned...

Regards
Guido Vollbeding
Organizer Independent JPEG Group