Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

JPEG 2000: which camera will support it? When?

42 views
Skip to first unread message

John Faughnan

unread,
Jan 28, 2004, 3:32:02 PM1/28/04
to
There was a good discussion in 2001 on when cameras would support
JPEG2000 [1]. Of course the expectation was late 2001 or early 2002.
Also in 2001, Analog Devices was supposed to have a chip out.[2]

Well, it's 3 years later. QuickTime 6x supports JPEG 2000 (iPhoto thus
displays it). Photoshop plug-ins shipped in Feb. 2003.
GraphicConverter (Mac) has two JPEG 2000 converters. The standard is
well defined and the licensing has been set.

The current Canon 6MPixel CMOS DIGIC sensor is all the sensor I need
for my next camera (to replace my much loved G2). The other things I
need to justify the buy are faster data paths in the camer, a (extra
bucks) 4GB Hitachi microdrive, USB 2.0/Firewire outputs, and .... JPEG
2000.

I figure I should be able to get all of the above in the fall of 2004
for somewhere between $500 to $800 US. Am I deluded by wishful
thinking? Do we think JPEG 2000 (both lossy and lossless) on the
camera is on the way soon?

john
jfau...@spamcop.net
www.faughnan.com/iphoto

meta: jfaughnan, jgfaughnan, digital imaging, digital camera, JPEG2K,
JPEG2000, adoption, implementation, technology.

[1] http://groups.google.com/groups?hl=en&lr=lang_en&ie=UTF-8&oe=UTF-8&threadm=ijp7ktkl6rs8tsaltqomsrrsif7elbbu2p%404ax.com&rnum=5

[2] http://www.dpreview.com/news/0102/01020601analogdevjpeg2000.asp
Analog Devices Develops IC for JPEG-2000
Analog Devices, Inc is developing a high-efficiency coding/decoding
integrated circuit (IC) to support the Joint Photographic Coding
Experts Group (JPEG) 2000, the next-generation international standard
for image compression. Sample shipments are scheduled to start in the
second quarter of this year, with chips for digital still cameras and
printers marking the firm's entry into the image processing IC sector.

Part-1 of the JPEG-2000 standard, the core portion defining
encoding/decoding processing, was adopted as an international standard
in December 2000. Digital still camera manufacturers are watching
developments closely, recognizing that the new standard will replace
existing JPEG technology. However, as the JPEG-2000 does not offer
compatibility with the existing standard, manufacturers hesitate to
adopt the format.

The company is developing two types of ICs. The first will implement
compression and decompression of image data under JPEG-2000,
incorporating the wavelet conversion and entropy encoding circuits
required. The more highly integrated second IC will single-chip the
ARM7TDMI 32-bit reduced instruction set computer (RISC) microprocessor
core from ARM Ltd of the UK, together with an interface and the first
chip circuits. The firm had commercialized an image compression chip
using wavelet conversion before commencing work on JPEG-2000.

Guido Vollbeding

unread,
Jan 29, 2004, 4:14:55 AM1/29/04
to
John Faughnan wrote:
>
> Do we think JPEG 2000 (both lossy and lossless) on the
> camera is on the way soon?

Forget JPEG 2000!
It was just marketing hype.
Standard JPEG is fine and will improve - JPEG 2000 is obsolete.

The next JPEG will probably be JPEG+ (JPEG plus; .jpp) - unlike
JPEG 2000 it will be backward compatible with today standard JPEG.

Regards
Guido

David Dyer-Bennet

unread,
Jan 29, 2004, 2:18:00 PM1/29/04
to
jfau...@spamcop.net (John Faughnan) writes:

> There was a good discussion in 2001 on when cameras would support
> JPEG2000 [1]. Of course the expectation was late 2001 or early 2002.
> Also in 2001, Analog Devices was supposed to have a chip out.[2]

I don't really see why I'd care about jpeg-2000. At least, from my
playing and what others report, there don't seem to be any particular
benefits. Why is it important to you?
--
David Dyer-Bennet, <mailto:dd...@dd-b.net>, <http://www.dd-b.net/dd-b/>
RKBA: <http://noguns-nomoney.com> <http://www.dd-b.net/carry/>
Photos: <dd-b.lighthunters.net> Snapshots: <www.dd-b.net/dd-b/SnapshotAlbum/>
Dragaera/Steven Brust: <http://dragaera.info/>

David J Taylor

unread,
Jan 29, 2004, 2:37:20 PM1/29/04
to

"David Dyer-Bennet" <dd...@dd-b.net> wrote in message
news:m2y8rq7...@gw.dd-b.net...
[]

> I don't really see why I'd care about jpeg-2000. At least, from my
> playing and what others report, there don't seem to be any particular
> benefits. Why is it important to you?
> --
> David Dyer-Bennet, <mailto:dd...@dd-b.net>, <http://www.dd-b.net/dd-b/>

12-bit JPEG might be nice, though! Most of the benfits of raw without the
size drawbacks.....

David


Ron Hunter

unread,
Jan 29, 2004, 6:17:03 PM1/29/04
to
David Dyer-Bennet wrote:

> jfau...@spamcop.net (John Faughnan) writes:
>
>
>>There was a good discussion in 2001 on when cameras would support
>>JPEG2000 [1]. Of course the expectation was late 2001 or early 2002.
>>Also in 2001, Analog Devices was supposed to have a chip out.[2]
>
>
> I don't really see why I'd care about jpeg-2000. At least, from my
> playing and what others report, there don't seem to be any particular
> benefits. Why is it important to you?

It does have many benefits, such as better compression, and better
detail retention with less artifacting. However, it appears that it has
been 'up the flagpole' for some time with very little in the way of
salutes.

David Dyer-Bennet

unread,
Jan 29, 2004, 7:04:27 PM1/29/04
to
"David J Taylor" <david-...@blueyonder.co.uk.not-this-bit> writes:

> "David Dyer-Bennet" <dd...@dd-b.net> wrote in message
> news:m2y8rq7...@gw.dd-b.net...
> []
>> I don't really see why I'd care about jpeg-2000. At least, from my
>> playing and what others report, there don't seem to be any particular
>> benefits. Why is it important to you?

> 12-bit JPEG might be nice, though! Most of the benfits of raw without the
> size drawbacks.....

Oops, yes, forgot that. I've seen other unofficial jpeg extensions to
handle it, and forgot it was official in 2000.


--
David Dyer-Bennet, <mailto:dd...@dd-b.net>, <http://www.dd-b.net/dd-b/>

David Dyer-Bennet

unread,
Jan 29, 2004, 7:04:58 PM1/29/04
to
Ron Hunter <rphu...@charter.net> writes:

Well, that's just it; those benefits don't appear in the reports I see
on it, or in my own little testing (I've got a number of programs that
support it).

Ron Hunter

unread,
Jan 29, 2004, 9:15:13 PM1/29/04
to
David Dyer-Bennet wrote:
> Ron Hunter <rphu...@charter.net> writes:
>
>
>>David Dyer-Bennet wrote:
>>
>>
>>>jfau...@spamcop.net (John Faughnan) writes:
>>>
>>>
>>>>There was a good discussion in 2001 on when cameras would support
>>>>JPEG2000 [1]. Of course the expectation was late 2001 or early 2002.
>>>>Also in 2001, Analog Devices was supposed to have a chip out.[2]
>>>
>>>I don't really see why I'd care about jpeg-2000. At least, from my
>>>playing and what others report, there don't seem to be any particular
>>>benefits. Why is it important to you?
>>
>>It does have many benefits, such as better compression, and better
>>detail retention with less artifacting. However, it appears that it
>>has been 'up the flagpole' for some time with very little in the way
>>of salutes.
>
>
> Well, that's just it; those benefits don't appear in the reports I see
> on it, or in my own little testing (I've got a number of programs that
> support it).

I have tested it somewhat and found that there are substantially less
artifacts for a given compression level.

John Faughnan

unread,
Jan 29, 2004, 10:18:53 PM1/29/04
to
Ron Hunter <rphu...@charter.net> wrote in message news:<101j53h...@corp.supernews.com>...

JPEG 2000 is the only potential standard I know of for lossless
compression. It's more efficient than TIFF and it's a much better
defined standard -- hence likely to be interoperable. The method of
lossy compression provides much better retention of data with less
artifact with smaller images than JPEG. It provides a wider dynamic
range than JPEG.

It's a real improvement. I'd like to see it.

john

David J Taylor

unread,
Jan 30, 2004, 2:34:10 AM1/30/04
to
"John Faughnan" <jfau...@spamcop.net> wrote in message
news:5c0dbfb4.04012...@posting.google.com...
[]

> JPEG 2000 is the only potential standard I know of for lossless
> compression. It's more efficient than TIFF and it's a much better
> defined standard -- hence likely to be interoperable.

PNG offers all that, doesn't it?

Cheers,
David


Zol.

unread,
Jan 30, 2004, 2:56:56 AM1/30/04
to
You can convert any Canon CRW file to jpeg2000 but why you should want to do that I have no
idea as the format irself isn`t widely used and has limited appeal compared with other
formats out there


"John Faughnan" <jfau...@spamcop.net> wrote in message
news:5c0dbfb4.04012...@posting.google.com...

Martin Brown

unread,
Jan 30, 2004, 4:06:27 AM1/30/04
to
In message <QhdSb.947$Gg6.9...@news-text.cableinet.net>, David J
Taylor <david-...@blueyonder.co.uk.not-this-bit> writes
>

12bit classical JPEG is already available. It just isn't used in
consumer goods.

Regards,
--
Martin Brown

John Faughnan

unread,
Jan 30, 2004, 9:13:14 AM1/30/04
to
"Zol." <Zol.@The_End_of_The_World.com> wrote in message news:<c7oSb.21149$pD4....@news-lhr.blueyonder.co.uk>...

> You can convert any Canon CRW file to jpeg2000 but why you should want to do that I have no
> idea as the format irself isn`t widely used and has limited appeal compared with other
> formats out therei

I think it's not true that there are better choices now in use. Maybe
in the lab.

This thread has convinced me, however, that JPEG2000 desperately needs
a marketing director. It does sound like a lot of knowledgeable photo
people are not familiar with it. In the absence of market interest it
is unlikely to be adopted.

If I were the licensing bodies involved with JPEG2000 I'd be
reexamining my strategy. For example, they may need to reduce or
eliminate any of the current licensing fees on software
encoder/decoders and look to make money only from hardware encoders.

Based on the limited market knowledge or interest in JPEG2000 I won't
look for it in a camera next year.

john
jfau...@spamcop.net
www.faughnan.com/iphoto

meta: jfaughnan, jgfaughnan, digital imaging, digital camera, JPEG2K,

JPEG2000, JPEG 2000, adoption, implementation, technology

Guido Vollbeding

unread,
Jan 30, 2004, 9:27:02 AM1/30/04
to
John Faughnan wrote:
>
> This thread has convinced me, however, that JPEG2000 desperately needs
> a marketing director. It does sound like a lot of knowledgeable photo
> people are not familiar with it. In the absence of market interest it
> is unlikely to be adopted.

Hmm, was I not clear enough?
All what JPEG2000 needs is to be ignored!
It is technically inferior to the current JPEG standard and its
potential capabilities, which are far from being exploited today.

Regards
Guido

David J Taylor

unread,
Jan 30, 2004, 9:51:23 AM1/30/04
to
> >12-bit JPEG might be nice, though! Most of the benfits of raw without
the
> >size drawbacks.....
>
> 12bit classical JPEG is already available. It just isn't used in
> consumer goods.

Yes, that was my point (deliberately understated) that you don't need to
go to JPEG 2000 to get 12-bits. I am actually using a 12-bit JPEG decode
in one of my applications....

Cheers,
David


KBob

unread,
Jan 30, 2004, 12:31:49 PM1/30/04
to
On Thu, 29 Jan 2004 18:04:58 -0600, David Dyer-Bennet <dd...@dd-b.net>
wrote:

>Ron Hunter <rphu...@charter.net> writes:
>
>> David Dyer-Bennet wrote:
>>
>>> jfau...@spamcop.net (John Faughnan) writes:
>>>
>>>>There was a good discussion in 2001 on when cameras would support
>>>>JPEG2000 [1]. Of course the expectation was late 2001 or early 2002.
>>>>Also in 2001, Analog Devices was supposed to have a chip out.[2]
>>> I don't really see why I'd care about jpeg-2000. At least, from my
>>> playing and what others report, there don't seem to be any particular
>>> benefits. Why is it important to you?
>>
>> It does have many benefits, such as better compression, and better
>> detail retention with less artifacting. However, it appears that it
>> has been 'up the flagpole' for some time with very little in the way
>> of salutes.
>
>Well, that's just it; those benefits don't appear in the reports I see
>on it, or in my own little testing (I've got a number of programs that
>support it).

I tried it but found that it did little except to trade the "hard"
JPEG artifacts for image softening, no real advantage IMHO.

Brad

unread,
Jan 30, 2004, 1:57:29 PM1/30/04
to
In article <401A69B6...@jpegclub.org>, gu...@jpegclub.org said...

You probably weren't clear enough because you understand neither JPEG2000
nor English.

Michael Schnell

unread,
Jan 30, 2004, 5:29:34 PM1/30/04
to
>Standard JPEG is fine and will improve - JPEG 2000 is obsolete.

In my tests JPEG 2000 gave better looking pictures at a given file size.
Moreover JPEG 2000 can do more than 8 Bit (Maybe JPG can do this, too, but I
did not see a converter that does it).

When capturing RAW, Ill be saving the images in JP2 / 12 bit after some
initial tweaking.

-Michael


Ron Hunter

unread,
Jan 30, 2004, 8:03:28 PM1/30/04
to

Total nonsense. However, it IS being ignored, and I don't see that
changing, so the question is moot.

Guido Vollbeding

unread,
Feb 2, 2004, 4:00:06 AM2/2/04
to
Michael Schnell wrote:
>
> >Standard JPEG is fine and will improve - JPEG 2000 is obsolete.
>
> In my tests JPEG 2000 gave better looking pictures at a given file size.

Because you probably didn't use appropriate JPEG compression settings.
For example, JPEG-2000 uses arithmetic coding, so you must also use JPEG
with arithmetic coding for comparision, otherwise you are comparing apples
with oranges.

> Moreover JPEG 2000 can do more than 8 Bit (Maybe JPG can do this, too, but I
> did not see a converter that does it).

Standard JPEG can also do 12 bits per pixel lossy and 16 bits per pixel lossless.

> When capturing RAW, Ill be saving the images in JP2 / 12 bit after some
> initial tweaking.

RAW from a Bayer 'image'? Well, that is a good match, because both Bayer
RAW and JPEG2000 are obsolete.

Regards
Guido

Guido Vollbeding

unread,
Feb 2, 2004, 4:15:42 AM2/2/04
to
Brad wrote:
>
> > > This thread has convinced me, however, that JPEG2000 desperately needs
> > > a marketing director. It does sound like a lot of knowledgeable photo
> > > people are not familiar with it. In the absence of market interest it
> > > is unlikely to be adopted.
> >
> > Hmm, was I not clear enough?
> > All what JPEG2000 needs is to be ignored!
> > It is technically inferior to the current JPEG standard and its
> > potential capabilities, which are far from being exploited today.
>
> You probably weren't clear enough because you understand neither JPEG2000
> nor English.

I'm just using English with artistic latitude ;-).
I have studied JPEG2000 enough to see that it is inferior.
Most people don't understand JPEG, being lean and simple in comparison.
It is not necessary to understand the cumbersome and inferior JPEG2000.
Better learn to understand JPEG, that's my advice.
In fact, I only recently discovered the major reason of JPEG's
suitability - even the original JPEG authors, not to mention the
JPEG2000 authors, didn't know the core JPEG property.

Regards
Guido

Ron Hunter

unread,
Feb 2, 2004, 10:48:43 AM2/2/04
to

I don't believe the word 'Bayer' appeared in his post. So you inserted
it to make your point? This is not ethical debate tactics.

Guido Vollbeding

unread,
Feb 2, 2004, 11:13:31 AM2/2/04
to
Ron Hunter wrote:
>
> I don't believe the word 'Bayer' appeared in his post. So you inserted
> it to make your point? This is not ethical debate tactics.

With some research you will find that the poster did not use the only
available non-Bayer (or better non-mosaic) RAW camera (Sigma/Foveon).
You will find that he used a Nikon D100.

So yes, I made the point that using one obscure technology (capturing
Bayer RAW) with another obscure technology (J2K compression) might be
a good match. I recommend to avoid both.

Regards
Guido

Ron Hunter

unread,
Feb 2, 2004, 8:08:59 PM2/2/04
to

RAW is RAW, it is the best the sensor can deliver, unprocessed. Seems
like a good place to start if you want the best image possible from a
given camera.
Are you sure you don't see George when you look in the mirror?

Guido Vollbeding

unread,
Feb 3, 2004, 3:33:55 AM2/3/04
to
Ron Hunter wrote:
>
> RAW is RAW, it is the best the sensor can deliver, unprocessed.

RAW from a Bayer camera is far away from a natural photograph,
missing 2/3s of the natural color spectrum!

Regards
Guido

Ron Hunter

unread,
Feb 3, 2004, 4:09:36 AM2/3/04
to

If it is the best the sensor has to offer, it has to do.
Note that it is modelled after the way the human (primate) eye works, so
can we really SEE any difference, given the 'hardware' of the human eye?
It's rather like spending money on a stereo to get flat response from
20 to 22,000hz, when one can't hear past 7,500hz. Why entertain the dog?

Guido Vollbeding

unread,
Feb 3, 2004, 4:34:00 AM2/3/04
to
Ron Hunter wrote:
>
> > RAW from a Bayer camera is far away from a natural photograph,
> > missing 2/3s of the natural color spectrum!
>
> If it is the best the sensor has to offer, it has to do.
> Note that it is modelled after the way the human (primate) eye works, so
> can we really SEE any difference, given the 'hardware' of the human eye?
> It's rather like spending money on a stereo to get flat response from
> 20 to 22,000hz, when one can't hear past 7,500hz. Why entertain the dog?

The human eye is different and does NOT work like the Bayer sensor!
My eye certainly doesn't like the Bayer look.

Regards
Guido

Lionel

unread,
Feb 3, 2004, 4:34:32 AM2/3/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

It's annoying enough to have that idiot Preddy jumping into every thread
he can find to advertise the SD9. How about you fuck off before you get
the same reputation? I'm *this* close to writing to Foveon & Sigma to
complain about you both spamming advertising material into this
newsgroup.

--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------

Lionel

unread,
Feb 3, 2004, 4:41:19 AM2/3/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

>Ron Hunter wrote:

It does, actually. Hit the library & read up on how the human retina
works. It's far more similar to Bayer sensors work & totally unlike the
way that Foveon sensors (or film, for that matter) work.

>My eye certainly doesn't like the Bayer look.

You & George like what your sugar-daddies at Sigma tell you to like.

Guido Vollbeding

unread,
Feb 3, 2004, 6:10:06 AM2/3/04
to
Lionel wrote:
>
> >RAW from a Bayer camera is far away from a natural photograph,
> >missing 2/3s of the natural color spectrum!
>
> It's annoying enough to have that idiot Preddy jumping into every thread
> he can find to advertise the SD9. How about you fuck off before you get
> the same reputation? I'm *this* close to writing to Foveon & Sigma to
> complain about you both spamming advertising material into this
> newsgroup.

Stay on-topic, please!
This thread is about *image data compression technology*!
Using a Bayer camera, capturing only 1/3 of the picture information
and then *adding* 2/3s of data by calculation *hurts* the image
compression purpose considerably!
Doing something like that is just a brain-damaged approach!
That's why it has to be denied to use Bayer sensing when talking
about appropriate image data compression!

Regards
Guido

David J Taylor

unread,
Feb 3, 2004, 6:29:49 AM2/3/04
to
> Stay on-topic, please!
> This thread is about *image data compression technology*!
> Using a Bayer camera, capturing only 1/3 of the picture information
> and then *adding* 2/3s of data by calculation *hurts* the image
> compression purpose considerably!
> Doing something like that is just a brain-damaged approach!
> That's why it has to be denied to use Bayer sensing when talking
> about appropriate image data compression!

Guido,

Turning it round, then, considering that the majority of cameras today
_do_ use Bayer mosaics, how does that affect what compression should be
used? I mean, what compression would be more suitable than JPEG if a
certain file size is required, or can one do a reasonable job simply by
tuning the parameters of JPEG such as chroma sub-sampling?

Cheers,
David


Lionel

unread,
Feb 3, 2004, 6:36:13 AM2/3/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

>Lionel wrote:


>>
>> >RAW from a Bayer camera is far away from a natural photograph,
>> >missing 2/3s of the natural color spectrum!
>>
>> It's annoying enough to have that idiot Preddy jumping into every thread
>> he can find to advertise the SD9. How about you fuck off before you get
>> the same reputation? I'm *this* close to writing to Foveon & Sigma to
>> complain about you both spamming advertising material into this
>> newsgroup.

Okay, don't say that I didn't warn you.

>Stay on-topic, please!
>This thread is about *image data compression technology*!

No, it's about support for JPEG2000. It's unrelated to any particular
image sensor technology.

>Using a Bayer camera, capturing only 1/3 of the picture information
>and then *adding* 2/3s of data by calculation *hurts* the image
>compression purpose considerably!
>Doing something like that is just a brain-damaged approach!
>That's why it has to be denied to use Bayer sensing when talking
>about appropriate image data compression!

You're full of shit. JPEG compression works well for the same reason
that Bayer sensors work well - both match the way that the human eye
processes chroma & luma information.

Guido Vollbeding

unread,
Feb 3, 2004, 7:26:21 AM2/3/04
to
> Turning it round, then, considering that the majority of cameras today
> _do_ use Bayer mosaics, how does that affect what compression should be
> used? I mean, what compression would be more suitable than JPEG if a
> certain file size is required, or can one do a reasonable job simply by
> tuning the parameters of JPEG such as chroma sub-sampling?

David,

JPEG and other image compression and image processing schemes are based
upon the basic digital imaging model. A key property of this digital
imaging model is the full-color-spectrum-per-pixel assumption.
Noticeably violating this basic assumption in the case of Bayer sensing
renders all further results from the usual model useless.
If I would see any opportunity to produce useful results from a Bayer
capture, I would let you know, honestly.
But my perception is that there is no such opportunity, so whatever you
do to compress/process a Bayer 'image', you will not get something near
a natural photograph experience.

If you are in the business to define an appropriate representation of
natural photographs in a digital form, you must start from the closest
approximation of such image. The usual true-color digital image is such
close approximation, but the Bayer 'image' definitely is not, because it
already lost a considerable amount of picture information. No digital
image compression expert would come to the idea to reduce the amount of
picture data in the Bayer pattern way - the Bayer approach is nothing
more than a technical artifact in times when people tried to capture
color images with monochrome image sensors. Nice try, but now we see
that it failed, and we need native (true-)color sensors.

Now here is a rough description about my recent JPEG discovery in this
relation:
The fundamental property of lossy image compression is the similarity
of different resolutions of the same image. "Lossy" compression means
that we assign *the same* output representation to *multiple*, *similar*
input representations. The basic similarity relation for images is
resolution, or scale, invariance: If we see the same image in different
resolutions (scales, sizes), or the same subject from different distances,
we talk about *the same* image (or subject).
Resolution is, beside the true-color-per-pixel property, one part of the
basic digital imaging model. JPEG is the optimal algorithm for lossy
image compression, because the DCT (Discrete Cosine Transform - the core
of JPEG) provides the best resolution separation property for digital
images. Wavelet transforms, as used in JPEG2000, for example, do *not*
provide such optimal resolution separation. See also chapter 4 of my
paper at http://jpegclub.org/temp/ .

Regards
Guido

David J Taylor

unread,
Feb 3, 2004, 9:00:32 AM2/3/04
to
> David,
>
> JPEG and other image compression and image processing schemes are based
> upon the basic digital imaging model. A key property of this digital
> imaging model is the full-color-spectrum-per-pixel assumption.
> Noticeably violating this basic assumption in the case of Bayer sensing
> renders all further results from the usual model useless.
[]
> Regards
> Guido

Whilst I hear what you say, I don't perceive it that way.

To me, JPEG can comprise a high-resolution monochrome channel, together
with lower resolution colour channels. Under those circumstances, it
seems to me idea for the Bayer sensor with its differing luminance and
chrominance resolutions.

Cheers,
David


Ron Hunter

unread,
Feb 3, 2004, 9:05:38 AM2/3/04
to

No, it has NOTHING to do with compression. The compression of a 3 color
sensor would work exactly the same as for a Bayer sensor. You are
trying to push a sensor technology. Stay on subject, or start another
thread.

Guido Vollbeding

unread,
Feb 3, 2004, 9:16:47 AM2/3/04
to
David J Taylor wrote:
>
> Whilst I hear what you say, I don't perceive it that way.
>
> To me, JPEG can comprise a high-resolution monochrome channel, together
> with lower resolution colour channels. Under those circumstances, it
> seems to me idea for the Bayer sensor with its differing luminance and
> chrominance resolutions.

Where does the Bayer sensor separates or differs luminance and
chrominance resolution?
I constantly hear this argument, but I can't perceive it that way ;-).
Look at the Bayer sensor or RAW format, please! Where do you have a
proper luminance information there which would allow for capturing a
"high-resolution monochrome channel"? It is just not there - all
what you have are the Red, Green, or Blue elements - no proper
luminance anywhere.

So while JPEG can comprise a high-resolution monochrome channel,
the Bayer scheme *prevents* the capture of a high-resolution
monochrome channel. You cannot take away the color masks
from the Bayer sensor even if you wanted just a B/W capture.

Regards
Guido

David J Taylor

unread,
Feb 3, 2004, 10:03:55 AM2/3/04
to
"Guido Vollbeding" <gu...@jpegclub.org> wrote in message
news:401FAD4F...@jpegclub.org...
[]

> So while JPEG can comprise a high-resolution monochrome channel,
> the Bayer scheme *prevents* the capture of a high-resolution
> monochrome channel. You cannot take away the color masks
> from the Bayer sensor even if you wanted just a B/W capture.
>
> Regards
> Guido

Yes, I see what you mean - at best it's an approximation to "luminance",
with some added noise.... It's amazing how well it works, then!

Cheers,
David


Guido Vollbeding

unread,
Feb 3, 2004, 10:14:46 AM2/3/04
to
David J Taylor wrote:
>
> Yes, I see what you mean - at best it's an approximation to "luminance",
> with some added noise.... It's amazing how well it works, then!

The JPEG luminance formula is as follows:

Y = 0.299 * R + 0.587 * G + 0.114 * B

If you have just *one* of the RGB elements as in the Bayer case,
you can see how well this "approximation to luminance" works, then...

Regards
Guido

Guido Vollbeding

unread,
Feb 3, 2004, 10:38:41 AM2/3/04
to
> The JPEG luminance formula is as follows:
>
> Y = 0.299 * R + 0.587 * G + 0.114 * B

BTW, from this formula you can see why the Bayer pattern contains 2
Green elements versus 1 Red and 1 Blue: Green contributes the most
part to the luminance. But there's still a lot missing...

Regards
Guido

David J Taylor

unread,
Feb 3, 2004, 10:56:36 AM2/3/04
to

"Guido Vollbeding" <gu...@jpegclub.org> wrote in message
news:401FBAE6...@jpegclub.org...
[]

> The JPEG luminance formula is as follows:
>
> Y = 0.299 * R + 0.587 * G + 0.114 * B
>
> If you have just *one* of the RGB elements as in the Bayer case,
> you can see how well this "approximation to luminance" works, then...
>
> Regards
> Guido

Yes, if one considers the individual pixels. With the right anti-alias
filtering, you can derive a good RGB from a quad of four, though. Then
you could apply standard JPEG techniques.

Cheers,
David


Alfred Molon

unread,
Feb 3, 2004, 1:05:07 PM2/3/04
to
In article <ehqu10pf789j48sg8...@4ax.com>, n...@alt.net
says...

> >RAW from a Bayer camera is far away from a natural photograph,
> >missing 2/3s of the natural color spectrum!
>
> It's annoying enough to have that idiot Preddy jumping into every thread
> he can find to advertise the SD9. How about you fuck off before you get
> the same reputation? I'm *this* close to writing to Foveon & Sigma to
> complain about you both spamming advertising material into this
> newsgroup.

It's quite obvious that capturing three colours per pixel is better than
capturing just one. That's BTW also how film works (three overlapping
colour layers).

--

Alfred Molon
------------------------------
http://groups.yahoo.com/group/Olympus_405060/
Olympus 5050 resource - http://www.molon.de/5050.html
Olympus 5060 resource - http://www.molon.de/5060.html

Lionel

unread,
Feb 3, 2004, 2:58:47 PM2/3/04
to
Kibo informs me that Alfred Molon <alfred_mo...@yahoo.com> stated
that:

>In article <ehqu10pf789j48sg8...@4ax.com>, n...@alt.net
>says...
>> >RAW from a Bayer camera is far away from a natural photograph,
>> >missing 2/3s of the natural color spectrum!
>>
>> It's annoying enough to have that idiot Preddy jumping into every thread
>> he can find to advertise the SD9. How about you fuck off before you get
>> the same reputation? I'm *this* close to writing to Foveon & Sigma to
>> complain about you both spamming advertising material into this
>> newsgroup.
>
>It's quite obvious that capturing three colours per pixel is better than
>capturing just one.

If it's done accurately, & at the same resolution, sure. If foveon comes
up with a sensor like that, & puts it in a decent camera body, I'll be
saving my pennies to buy one myself.

> That's BTW also how film works (three overlapping
>colour layers).

True, but it's not how eyes work. There's a reason why every lossy image
& video format preserves luma detail, but throws away lots of chroma
detail - it's because eyes just don't need as much colour detail as they
do brightness detail.

David J. Littleboy

unread,
Feb 3, 2004, 4:41:07 PM2/3/04
to

"David J Taylor" <david-...@blueyonder.co.uk.not-this-bit> wrote in
message news:UwPTb.5381$7q1.49...@news-text.cableinet.net...

That formula has nothing to do with Bayer: it reflects human visual
perception and applies to any RGB image. And talking about Bayer in terms of
quads of pixels is simply wrong.

The two of you have gotten yourselves confused.

The problem at hand (compressing RAW images) should be handled by
_independently_ and losslessly compressing (a) the set of R measurements as
a monochrome image, (b) the set of G measurements as a monochrome image, and
(c) the set of B measurements as a monochrome image. That set of three
compressed images can then be independently reconstructed so that one can
then apply better (more computationally intesive) Bayer demosaicing
algorithms outside the camera.

So the relevant question for the RAW problem is what are the best lossless
monochrome compression techniques. Since RAW formats are proprietary, this
has nothing to do with JPEG vs. JPEG2000.

If you do your Bayer demosaicing in camera, then you've got an RGB matrix to
which either JPEG or JPWG2000 can be applied, and which works better is
going to be independent of Bayer demosaicing.

David J. Littleboy
Tokyo, Japan


David J Taylor

unread,
Feb 3, 2004, 5:32:26 PM2/3/04
to
"David J. Littleboy" <dav...@gol.com> wrote in message
news:bvp2ds$h7h$1...@nnrp.gol.com...
[]

> So the relevant question for the RAW problem is what are the best
lossless
> monochrome compression techniques. Since RAW formats are proprietary,
this
> has nothing to do with JPEG vs. JPEG2000.
>
> If you do your Bayer demosaicing in camera, then you've got an RGB
matrix to
> which either JPEG or JPWG2000 can be applied, and which works better is
> going to be independent of Bayer demosaicing.

David, thanks for your input. I must confess to thinking more about the
composite RGB image, as I don't use RAW myself. I can see Guido's point,
though, as JPEG typically implies separate luminance and chrominance
sampling, and that perhaps the statistics of the luminance may differ
according to the sensor type.

Cheers,
David


David J. Littleboy

unread,
Feb 3, 2004, 5:44:33 PM2/3/04
to

"David J Taylor" <david-...@blueyonder.co.uk.not-this-bit> wrote:

> "David J. Littleboy" <dav...@gol.com> wrote:
> > So the relevant question for the RAW problem is what are the best
> lossless
> > monochrome compression techniques. Since RAW formats are proprietary,
> this
> > has nothing to do with JPEG vs. JPEG2000.
> >
> > If you do your Bayer demosaicing in camera, then you've got an RGB
> matrix to
> > which either JPEG or JPWG2000 can be applied, and which works better is
> > going to be independent of Bayer demosaicing.
>
> David, thanks for your input. I must confess to thinking more about the
> composite RGB image, as I don't use RAW myself.

But you will at some point, most likely. Not having to commit to an
incorrect white balance in advance is a big advantage<g>.

> I can see Guido's point,
> though, as JPEG typically implies separate luminance and chrominance
> sampling, and that perhaps the statistics of the luminance may differ
> according to the sensor type.

Snore. It'll mean that images from a 3-color system compress less well, but
other than that, it's irrelevant. Guido's "point" isn't a point at all: it's
a stupid, technically incorrect, rant. Bayer is flipping amazing: you get as
much luminance information as can be correctly rendered in a matrix of that
size (without aliasing) and you get as much color as the human eye can
detect. There isn't anything missing that you can see or use.

And anyway, since the only cameras that actually measure 3 colors everywhere
are dogs, they're irrelevant.

David J Taylor

unread,
Feb 3, 2004, 5:49:25 PM2/3/04
to
"David J. Littleboy" <dav...@gol.com> wrote in message
news:bvp5eb$hfl$1...@nnrp.gol.com...
[]

> > David, thanks for your input. I must confess to thinking more about
the
> > composite RGB image, as I don't use RAW myself.
>
> But you will at some point, most likely. Not having to commit to an
> incorrect white balance in advance is a big advantage<g>.

Maybe, but the Nikons I use seem to do quite a good job (for the type of
shots I take). Most of my presentation is individual images on a CRT or
LCD display rather than prints where you might view more than one image at
a time.

[]


> And anyway, since the only cameras that actually measure 3 colors
everywhere
> are dogs, they're irrelevant.

Let's concentrate on what the majority use and enjoy!

Cheers,
David


Hans-Georg Michna

unread,
Feb 4, 2004, 5:02:38 AM2/4/04
to
Guido,

while reading messages in a related thread, one question made me
curious. Do the pixel color filters in a Bayer sensor filter out
almost all off-color light or do they only reduce the other
colors somewhat but do not block them entirely?

I hope I formulated the question such that it can be understood.
In other words, does a green filtered pixel only measure green
light or does it only have a somewhat pronounced green
sensitivity but still a lot of general luminance information?

By the way, I think that these religious Foveon-Bayer wars tend
to miss the point a little. As long as the sensors comprise
single pixel sensors, the question is, what is the optimal color
filter distribution? And perhaps whether the pixels should be
arranged in squares or like bee-hives. I also wonder whether any
manufacturer has got the idea to have four pixels filtered not
as red-green-green-blue, but as white-red-green-blue for better
luminance information.

Hans-Georg

--
No mail, please.

Guido Vollbeding

unread,
Feb 4, 2004, 6:15:47 AM2/4/04
to
Hans-Georg Michna wrote:
>
> while reading messages in a related thread, one question made me
> curious. Do the pixel color filters in a Bayer sensor filter out
> almost all off-color light or do they only reduce the other
> colors somewhat but do not block them entirely?
>
> I hope I formulated the question such that it can be understood.
> In other words, does a green filtered pixel only measure green
> light or does it only have a somewhat pronounced green
> sensitivity but still a lot of general luminance information?

Color perception is a somewhat difficult science and sometimes
controversal matter.
According to tri-stimulus color theory we make up all colors
from only 3 basic (primary) colors. The primary colors have a
certain response function over the wavelenghts spectrum which
just peaks at a Red, Green, or Blue point. There are compromises
with advantages and disadvantages whether to make the response
curves more narrow or more wide. So it is not the full truth
if people complain about possibly wide response curves in the
Foveon sensor which would lead to bad color separation. A
certain wideness is necessary for appropriate output color gamut.
See the figures in slides nr. 33-37 in the Foveon presentation
to get an idea:

"Silicon as a Color Filter"
"Silicon Color Separation"
"X3 Spectral Response Curves"
"Human Cone Spectral Response Curves"
"Color-Matching Functions"

http://www-inst.eecs.berkeley.edu/~cs39j/fa03/session03a.html

> By the way, I think that these religious Foveon-Bayer wars tend
> to miss the point a little. As long as the sensors comprise
> single pixel sensors, the question is, what is the optimal color
> filter distribution? And perhaps whether the pixels should be
> arranged in squares or like bee-hives. I also wonder whether any
> manufacturer has got the idea to have four pixels filtered not
> as red-green-green-blue, but as white-red-green-blue for better
> luminance information.

The mosaic filter approach is an unsuccessful approach, whatever
you do. You must avoid that. (Your white element wouldn't help
due to poor resolution.) Foveon seems to be a succcessful
solution - another approach was taken in video with 3-chip design
and color-separating prism, but with high cost and other problems.

Regards
Guido

Mark Herring

unread,
Feb 4, 2004, 9:42:14 AM2/4/04
to
The idea is to match the response of the color filters in the eye. If
there is too much out of band response, there will be errors.

**************************
Mark Herring, Pasadena, Calif.
Private e-mail: Just say no to "No".

eawck...@yahoo.com

unread,
Feb 4, 2004, 1:09:58 PM2/4/04
to
Guido Vollbeding <gu...@jpegclub.org> wrote:

> The mosaic filter approach is an unsuccessful approach [...]

I asked you before: are you this stupid in real life, or is this just
some sort of act you are putting on?

> You must avoid that.

The CFA is evidence of what is called "engineering". While you
blather on metaphysically about "unsuccessful approaches" and things
we must "avoid" (Lord Kelvin (1895): "Heavier than air flying
machiners are impossible"), it nevertheless remains a fact that CFA
cameras produce completely acceptable (if not excellent) images
anyways.

"If physical reality says you are an ass, you are an ass." (Uncle Al)

> Foveon seems to be a succcessful solution [...]

Success is defined by something that currently produces _worse_ colour
than the so-called unsuccessful design?

Your thought processes are, well, difficult to discern.

> another approach was taken in video with 3-chip design
> and color-separating prism, but with high cost and other problems.

Still cameras have a splendid advantage over video: they can take a
few seconds to digest the data and produce a better image. When you
are given only a 1/30th of a second you have to use a different
solution or accept the consequences.

But thats more of that over-your-head engineering for you.
Quality/cost tradeoffs are made all the time in the real world: deal
with it.

eawck...@yahoo.com

unread,
Feb 4, 2004, 1:35:00 PM2/4/04
to
Hans-Georg Michna <hans-georgN...@michna.com> wrote:

> Do the pixel color filters in a Bayer sensor filter out
> almost all off-color light or do they only reduce the other
> colors somewhat but do not block them entirely?

See http://astrosurf.com/buil/us/digit/spectra.htm for measurements of
the Canon 10D's response. I would expect most cameras to be more or
less the same.

> In other words, does a green filtered pixel only measure green
> light or does it only have a somewhat pronounced green
> sensitivity but still a lot of general luminance information?

Note there is luminance information in all the channels of a CFA, not
just the green. The green is "assumed" to be the luminance simply
because there is more of it on the sensor (and, of course, because our
eyes are more sensitive to it).

However, I highly recommend perusing some demosaic papers because alot
of the blather in this newsgroup re: "luminance" is mainly a
red-herring: much, MUCH, more important is channel-to-channel
correlation. Without this the CFA approach would indeed be a poor
one. I've posted references in the past in this newsgroup.

> By the way, I think that these religious Foveon-Bayer wars tend
> to miss the point a little.

The Bayer-Foveon wars exist so we can laugh at idiots like George
Preddy and his company of sock puppets.

Chris Brown

unread,
Feb 4, 2004, 1:43:00 PM2/4/04
to
In article <d3c57e81.0402...@posting.google.com>,

<eawck...@yahoo.com> wrote:
>
>The CFA is evidence of what is called "engineering". While you
>blather on metaphysically about "unsuccessful approaches" and things
>we must "avoid" (Lord Kelvin (1895): "Heavier than air flying
>machiners are impossible"),

That one always amused me. Either he never noticed birds, or must have
decided it was impossible to engineer a power source with a sufficient power
to weight ratio to power something, say, the size of a swan.

Given that swans do a pretty good job of turning fish and bread into heavier
than air flight, he must have realised that he was on dodgy ground.

Oh yeah, and as for Guido, I don't think anyone really takes him seriously.
The large numbers of professional and amateur photographers getting
perfectly good results from mosaic-sensor digital cameras every day, many of
which are now rather superior to traditional wet-process smll format
photography, lowers his position to that of "ill founded rant".

>> Foveon seems to be a succcessful solution [...]
>
>Success is defined by something that currently produces _worse_ colour
>than the so-called unsuccessful design?
>
>Your thought processes are, well, difficult to discern.

I congratulate you on your tact under fire. :-)

Hans-Georg Michna

unread,
Feb 5, 2004, 2:39:15 AM2/5/04
to
Hans-Georg Michna <hans-georgN...@michna.com> wrote:

>... Do the pixel color filters in a Bayer sensor filter out


>almost all off-color light or do they only reduce the other
>colors somewhat but do not block them entirely?
>
>I hope I formulated the question such that it can be understood.
>In other words, does a green filtered pixel only measure green
>light or does it only have a somewhat pronounced green
>sensitivity but still a lot of general luminance information?

Thank you, all, for the replies, but I guess I have to try to
reformulate the question, as all the answers so far didn't
answer my (single-bit) question.

Let me ask this way: Does the red filter (on the respective
pixels) in a Bayer sensor let only red light pass? Or is it a
pink filter that lets all colors pass, only more red than the
others?

Kevin McMurtrie

unread,
Feb 5, 2004, 3:49:02 AM2/5/04
to
In article <5c0dbfb4.04012...@posting.google.com>,
jfau...@spamcop.net (John Faughnan) wrote:

> There was a good discussion in 2001 on when cameras would support
> JPEG2000 [1]. Of course the expectation was late 2001 or early 2002.
> Also in 2001, Analog Devices was supposed to have a chip out.[2]
>
> Well, it's 3 years later. QuickTime 6x supports JPEG 2000 (iPhoto thus
> displays it). Photoshop plug-ins shipped in Feb. 2003.
> GraphicConverter (Mac) has two JPEG 2000 converters. The standard is
> well defined and the licensing has been set.
>
> The current Canon 6MPixel CMOS DIGIC sensor is all the sensor I need
> for my next camera (to replace my much loved G2). The other things I
> need to justify the buy are faster data paths in the camer, a (extra
> bucks) 4GB Hitachi microdrive, USB 2.0/Firewire outputs, and .... JPEG
> 2000.
>
> I figure I should be able to get all of the above in the fall of 2004
> for somewhere between $500 to $800 US. Am I deluded by wishful
> thinking? Do we think JPEG 2000 (both lossy and lossless) on the
> camera is on the way soon?
>
> john
> jfau...@spamcop.net
> www.faughnan.com/iphoto
>
[snip]

The quality to compression ratio of digicams could be improved
considerably by using dynamically generated Huffman dictionaries. This
is the "Baseline optimized" option in Photoshop that makes JPEGs smaller
by improving the final lossless stage of compression.

The problem is, creating a custom Huffman table requires another pass
over the JPEG data. The JPEG data can no longer be streamed through an
encoder chip that has the baseline Huffman table in ROM. It needs to be
buffered in lots of RAM for two pass processing. That wasn't practical
in the past but it should be standard soon.

Guido Vollbeding

unread,
Feb 5, 2004, 4:09:11 AM2/5/04
to
Kevin McMurtrie wrote:
>
> The quality to compression ratio of digicams could be improved
> considerably by using dynamically generated Huffman dictionaries. This
> is the "Baseline optimized" option in Photoshop that makes JPEGs smaller
> by improving the final lossless stage of compression.
>
> The problem is, creating a custom Huffman table requires another pass
> over the JPEG data. The JPEG data can no longer be streamed through an
> encoder chip that has the baseline Huffman table in ROM. It needs to be
> buffered in lots of RAM for two pass processing. That wasn't practical
> in the past but it should be standard soon.

I don't think it will be practical soon, but there's a better alternative.
Wait another 5 or so years, then they can apply the JPEG arithmetic coding
option (the patents expire then).
Arithmetic coding needs only a single processing pass, and it compresses
even better than the optimized Huffman option! You can try this today
with the Jpegcrop program, see http://jpegclub.org/jpegcrop/ for more
details.

BTW, JPEG2000 uses arithmetic coding by default, that's one reason why
industry doesn't adopt it, due to patent license requirements.
But J2K is inferior and obsolete anyway, so not a big deal there.

Regards
Guido

Guido Vollbeding

unread,
Feb 5, 2004, 4:33:09 AM2/5/04
to
Chris Brown wrote:
>
> Oh yeah, and as for Guido, I don't think anyone really takes him seriously.
> The large numbers of professional and amateur photographers getting
> perfectly good results from mosaic-sensor digital cameras every day, many of
> which are now rather superior to traditional wet-process smll format
> photography, lowers his position to that of "ill founded rant".

I don't expect to be taken seriously by incompetent people.
I'm too long in this business to be impressed by the fakes
and hoaxes from so-called 'experts', or by the acceptance
of crap by the crowd. I only hope that they don't bother
me too much with their junk.

Regards
Guido

Lionel

unread,
Feb 5, 2004, 4:48:21 AM2/5/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

>Chris Brown wrote:


>>
>> Oh yeah, and as for Guido, I don't think anyone really takes him seriously.
>> The large numbers of professional and amateur photographers getting
>> perfectly good results from mosaic-sensor digital cameras every day, many of
>> which are now rather superior to traditional wet-process smll format
>> photography, lowers his position to that of "ill founded rant".
>
>I don't expect to be taken seriously by incompetent people.

You & Geroge aren't taken seriously by /anyone/, competant or or
otherwise.

imbsysop

unread,
Feb 5, 2004, 5:00:04 AM2/5/04
to

wow .. and you were carved right out of the thighs of Zeus I presume ?

Martin Brown

unread,
Feb 5, 2004, 4:03:54 AM2/5/04
to
In message <43s320pbamhouj4ia...@4ax.com>, Hans-Georg
Michna <hans-georgN...@michna.com> writes

>Hans-Georg Michna <hans-georgN...@michna.com> wrote:
>
>>... Do the pixel color filters in a Bayer sensor filter out
>>almost all off-color light or do they only reduce the other
>>colors somewhat but do not block them entirely?

They are meant to do this in as far as it is physically possible.


>
>Thank you, all, for the replies, but I guess I have to try to
>reformulate the question, as all the answers so far didn't
>answer my (single-bit) question.
>
>Let me ask this way: Does the red filter (on the respective
>pixels) in a Bayer sensor let only red light pass? Or is it a
>pink filter that lets all colors pass, only more red than the
>others?

Usually the red filter is the only one with a really clean cutoff that
does what it says on the can. The green and blue filters may have some
feeble transmission in red or near IR. Easy enough to check - put a
Wratten 25 or better Wratten 29 filter in front of your camera and shoot
through that.

Bayer filters are meant to be nominally pure RGB colours (unless they
are the alternative more sensitive secondary pattern CMYG).

My old Kodak DC-120 used to go haywire shooting solar prominences in
monochromatic red hydrogen light at 656nm - all channels saw signal.

Regards,
--
Martin Brown

Azzz1588

unread,
Feb 5, 2004, 10:22:43 AM2/5/04
to
In article <40220DD5...@jpegclub.org>, Guido Vollbeding
<gu...@jpegclub.org> writes:

>I don't expect to be taken seriously


Have no fear, no one here ever takes you seriously............


"Only a Gentleman can insult me, and a true Gentleman never will..."


Michael Schnell

unread,
Feb 5, 2004, 12:08:11 PM2/5/04
to
>RAW from a Bayer camera is far away from a natural photograph,
>missing 2/3s of the natural color spectrum!
>

The human eye is missing 99,9999 % of the natural color spectrum (which is
continuos and not restricted to 3 "basic" colors) !

-Michael


Michael Schnell

unread,
Feb 5, 2004, 12:14:35 PM2/5/04
to
>Turning it round, then, considering that the majority of cameras today
>_do_ use Bayer mosaics, how does that affect what compression should be
>used?

Maybe it really does. If a camera (like the human eye) offers more
resolution on grayscale than on color, the compression could be more tight
on color than on grayscale. Or, regarding a Bayer sensor, tighter on red and
blue than on green.

AFAIK, JP2000 provides a color space and multiple differently compressed
monochrome pictures. So an optimum color space for the appropriate sensor
can be selected.

-Michael


Lionel

unread,
Feb 5, 2004, 12:58:07 PM2/5/04
to
Kibo informs me that Michael Schnell
<mschnell_at_b...@otmail.com> stated that:

>>Turning it round, then, considering that the majority of cameras today
>>_do_ use Bayer mosaics, how does that affect what compression should be
>>used?
>
>Maybe it really does. If a camera (like the human eye) offers more
>resolution on grayscale than on color, the compression could be more tight
>on color than on grayscale. Or, regarding a Bayer sensor, tighter on red and
>blue than on green.

Which is exactly what JPEG compression does, for exactly the same reason
that mosaic sensors distribute colours the way they do. In other words,
they both take advantage of the way the eye works.

David Dyer-Bennet

unread,
Feb 5, 2004, 1:59:43 PM2/5/04
to
Alfred Molon <alfred_mo...@yahoo.com> writes:

> In article <ehqu10pf789j48sg8...@4ax.com>, n...@alt.net
> says...
>> >RAW from a Bayer camera is far away from a natural photograph,
>> >missing 2/3s of the natural color spectrum!
>>
>> It's annoying enough to have that idiot Preddy jumping into every thread
>> he can find to advertise the SD9. How about you fuck off before you get
>> the same reputation? I'm *this* close to writing to Foveon & Sigma to
>> complain about you both spamming advertising material into this
>> newsgroup.
>
> It's quite obvious that capturing three colours per pixel is better than
> capturing just one. That's BTW also how film works (three overlapping
> colour layers).

It's "quite obvious" that that's true if all other things are equal.

It's *NOT* particularly obvious *how much* better. It's also not
obvious that all other things are in fact equal.

Foveon pictures don't look particularly better than Bayer pictures in
my experience so far. So I'm not willing to tie myself down to the
limitations of the only Foveon camera that I could possibly afford
(Sigma).
--
David Dyer-Bennet, <mailto:dd...@dd-b.net>, <http://www.dd-b.net/dd-b/>
RKBA: <http://noguns-nomoney.com> <http://www.dd-b.net/carry/>
Photos: <dd-b.lighthunters.net> Snapshots: <www.dd-b.net/dd-b/SnapshotAlbum/>
Dragaera/Steven Brust: <http://dragaera.info/>

Alfred Molon

unread,
Feb 5, 2004, 3:50:19 PM2/5/04
to
In article <154420hcfmpv0r81a...@4ax.com>, n...@alt.net
says...

> Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:
>
> >Chris Brown wrote:
> >>
> >> Oh yeah, and as for Guido, I don't think anyone really takes him seriously.
> >> The large numbers of professional and amateur photographers getting
> >> perfectly good results from mosaic-sensor digital cameras every day, many of
> >> which are now rather superior to traditional wet-process smll format
> >> photography, lowers his position to that of "ill founded rant".
> >
> >I don't expect to be taken seriously by incompetent people.
>
> You & Geroge aren't taken seriously by /anyone/, competant or or
> otherwise.

Speak for yourself.

jpc

unread,
Feb 5, 2004, 4:45:28 PM2/5/04
to
On Thu, 05 Feb 2004 08:39:15 +0100, Hans-Georg Michna
<hans-georgN...@michna.com> wrote:

>Hans-Georg Michna <hans-georgN...@michna.com> wrote:
>

>Thank you, all, for the replies, but I guess I have to try to
>reformulate the question, as all the answers so far didn't
>answer my (single-bit) question.
>
>Let me ask this way: Does the red filter (on the respective
>pixels) in a Bayer sensor let only red light pass? Or is it a
>pink filter that lets all colors pass, only more red than the
>others?
>

If you want to see how much of each color the Bayer filters on your
camera passes or blocks do the following.


Download the tif image of visible spectrum you find at
http://home.earthlink.net/~skywise711/LasersOptics/Reference/reference.html.
It is a reasonablely intense image with wavelength markers

It is also a long narrow image so I cropped and resampled it to print
it on a 4 by 6. You could also crop out the violet section beyond the
dark blue since that is only an artistic attempt at showing UV.

Print it out and take an image under diffuse sunlight.

Bring the image into a photo editor where you can display the
individual channels; red, green and blue. I used Photoshop but I
believe you could do this with any medium to high end image editor.

Duplicate each channel to make a new image and convert to greyscale.
Where it's bright you've collected colors, where it's black you've
blocked them.

Tile your images and compare. If you need numbers, download ImageJ so
you can plot out the relative intensities of each grey scale image.

With my camera the red channel completely blocked all blue but passed
yellow and some green, The green channel blocked the red, passed the
yellow and green plus more blue than I expected. The blue channel
passed the blue and after dropping quickly to about 10 percent slowly
fell without completely blocking either the red or green.

jpc


Lionel

unread,
Feb 5, 2004, 4:45:59 PM2/5/04
to
Kibo informs me that Alfred Molon <alfred_mo...@yahoo.com> stated
that:

>In article <154420hcfmpv0r81a...@4ax.com>, n...@alt.net

>says...
>> Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:
>>
>> >Chris Brown wrote:
>> >>
>> >> Oh yeah, and as for Guido, I don't think anyone really takes him seriously.
>> >> The large numbers of professional and amateur photographers getting
>> >> perfectly good results from mosaic-sensor digital cameras every day, many of
>> >> which are now rather superior to traditional wet-process smll format
>> >> photography, lowers his position to that of "ill founded rant".
>> >
>> >I don't expect to be taken seriously by incompetent people.
>>
>> You & Geroge aren't taken seriously by /anyone/, competant or or
>> otherwise.
>
>Speak for yourself.

You're one one of the "other loons" that I referred to a different post
on the same subject.

HTH, etc.

Alfred Molon

unread,
Feb 5, 2004, 4:47:42 PM2/5/04
to
In article <i6e5201fpn0vn4j9f...@4ax.com>, n...@alt.net
says...

> You're one one of the "other loons" that I referred to a different post
> on the same subject.

You just made it into my killfile.

Hans-Georg Michna

unread,
Feb 5, 2004, 5:40:25 PM2/5/04
to
Martin Brown <|||newspam|||@nezumi.demon.co.uk> wrote:

>Usually the red filter is the only one with a really clean cutoff that
>does what it says on the can. The green and blue filters may have some
>feeble transmission in red or near IR. Easy enough to check - put a
>Wratten 25 or better Wratten 29 filter in front of your camera and shoot
>through that.
>
>Bayer filters are meant to be nominally pure RGB colours (unless they
>are the alternative more sensitive secondary pattern CMYG).
>
>My old Kodak DC-120 used to go haywire shooting solar prominences in
>monochromatic red hydrogen light at 656nm - all channels saw signal.

Martin,

thanks a lot! As usual, things are more complex than one first
thinks.

Hans-Georg Michna

unread,
Feb 5, 2004, 5:40:28 PM2/5/04
to
jpc wrote:

>With my camera the red channel completely blocked all blue but passed
>yellow and some green, The green channel blocked the red, passed the
>yellow and green plus more blue than I expected. The blue channel
>passed the blue and after dropping quickly to about 10 percent slowly
>fell without completely blocking either the red or green.

Thanks for the good information! Interesting.

Lionel

unread,
Feb 5, 2004, 5:59:49 PM2/5/04
to
Kibo informs me that Alfred Molon <alfred_mo...@yahoo.com> stated
that:

>In article <i6e5201fpn0vn4j9f...@4ax.com>, n...@alt.net

>says...
>
>> You're one one of the "other loons" that I referred to a different post
>> on the same subject.
>
>You just made it into my killfile.

If preddy isn't already in there, that's just fine with me.

jpc

unread,
Feb 5, 2004, 10:12:48 PM2/5/04
to

Working out a way to measure the Bayer filter curves is something I've
been thinking about doing for a year or so. I was just curious how it
could be done cheaply and easily. And if you don't mind me asking why
are you interested?

jpc

Martin Brown

unread,
Feb 5, 2004, 5:09:04 PM2/5/04
to
In message <vqb520p6d2gm6q6oc...@4ax.com>, jpc@?.?.invalid
writes

>On Thu, 05 Feb 2004 08:39:15 +0100, Hans-Georg Michna
><hans-georgN...@michna.com> wrote:
>
>>Hans-Georg Michna <hans-georgN...@michna.com> wrote:
>>
>>Let me ask this way: Does the red filter (on the respective
>>pixels) in a Bayer sensor let only red light pass? Or is it a
>>pink filter that lets all colors pass, only more red than the
>>others?
>>
>
>If you want to see how much of each color the Bayer filters on your
>camera passes or blocks do the following.
>
>
>Download the tif image of visible spectrum you find at
>http://home.earthlink.net/~skywise711/LasersOptics/Reference/reference.html.
>It is a reasonablely intense image with wavelength markers
>
>It is also a long narrow image so I cropped and resampled it to print
>it on a 4 by 6. You could also crop out the violet section beyond the
>dark blue since that is only an artistic attempt at showing UV.
>
>Print it out and take an image under diffuse sunlight.

I hope you are joking.

This will only show you the reflectivity of the ink system used by your
printer when it tries to emulate the visual appearance of a solar
spectrum. And it won't even do that particularly well.

If you want to characterise a camera's filters the simplest test at a
consumer level is to photograph the reflection of a solar spectrum off a
silver shovelware CD. You will likely be very surprised by the result.

Cokin do a 2 point star diffraction grating that also does a reasonable
spectral dispersion for empirical testing of filters.

Regards,
--
Martin Brown

Guido Vollbeding

unread,
Feb 6, 2004, 3:27:47 AM2/6/04
to

Of course: The Bayer RAW is missing 2/3s of the natural color spectrum
*which the human eye is sensitive for*!

And tell that to the 'experts' here which always like to speak about
sampling and aliasing artifacts of the Foveon sensor:
The art of Bayer sensing is a *major* violation of basic sampling
assumptions, rendering all further results useless.
The Foveon sensor gives a *much* better approximation to native
photographs.

Regards
Guido

Lionel

unread,
Feb 6, 2004, 4:29:39 AM2/6/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

>Michael Schnell wrote:


>>
>> >RAW from a Bayer camera is far away from a natural photograph,
>> >missing 2/3s of the natural color spectrum!
>> >
>>
>> The human eye is missing 99,9999 % of the natural color spectrum (which is
>> continuos and not restricted to 3 "basic" colors) !
>
>Of course: The Bayer RAW is missing 2/3s of the natural color spectrum
>*which the human eye is sensitive for*!

Do you realise that your response related in almost no way to Michael's
comment? (Quite apart from it being untrue, of course.)

>And tell that to the 'experts' here which always like to speak about
>sampling and aliasing artifacts of the Foveon sensor:
>The art of Bayer sensing is a *major* violation of basic sampling
>assumptions

Which ones? - Please quote the actual rules concerned, eg; Nyquist, etc.

>, rendering all further results useless.

Then why does their output look so good?

>The Foveon sensor gives a *much* better approximation to native
>photographs.

Except that we don't /need/ a "much better approximation to native
photographs", because mosaic sensors pick up all the information that
the eye needs it to pick up. Sure, it'd be nice to have three colours at
every sensor pixel, & there are lots of nice things we could use that
extra data for, but we don't actually /need/ it. And if it comes at the
expense of a clunky camera, buggy colour & sub-average lenses, then no
thanks, I'll stick with something that I know can produce the results I
demand.

Guido Vollbeding

unread,
Feb 6, 2004, 5:07:33 AM2/6/04
to
Lionel wrote:
>
> >The art of Bayer sensing is a *major* violation of basic sampling
> >assumptions
>
> Which ones? - Please quote the actual rules concerned, eg; Nyquist, etc.

The basic assumption of sampling theory is that every single sample
sufficiently covers the range of susceptible values.
The analog-digital conversion must result in quantized values which
cover the susceptible range.
This digital quantization of values is, beside resolution (digital
quantization in space), one of the two basic sampling parameters.

If the quantization range of single values is as limited as in the
Bayer case, the other parameter, resolution in space of multiple
values, does not make sense. That's why the 'MP' rating of Bayer
sensors is wrong and misleading.

> I'll stick with something that I know can produce the results I demand.

Nice for you, but not for me...

Regards
Guido

Ron Hunter

unread,
Feb 6, 2004, 6:18:17 AM2/6/04
to
Guido Vollbeding wrote:

Now you are revealing your ignorance. The Bayer system is modelled
after the way in which primate vision works. Plonk.

Hans-Georg Michna

unread,
Feb 6, 2004, 6:58:59 AM2/6/04
to
jpc wrote:

jpc,

pure curiosity. I'm wondering how much emphasis they put on high
color resolution and precision.

The human eye has two different kind of interspersed sensors. It
uses high resolution color sensitive cells and a sprinkling of
black and white sensitive cells, which are much more sensitive
and are useful mainly at night. Interestingly we have no black &
white cells in the focus spot, so we can't focus very well on
anything at night. In fact, you may have to look elsewhere to
see something more clearly at night.

Maybe one day we will have digital cameras with two kinds of
interspersed sensors, one bunch for high quality, high light
situations and a thinner grid of high sensitivity, lower quality
sensors for low light situations. I would appreciate this and
gladly dispense with the flash.

I think there is at least one camera that already works that
way, but I have no details. I also remember that at least one
Sony camera has a special low light setting, but I don't know
whether it actually has different kinds of sensors.

jpc

unread,
Feb 6, 2004, 10:10:23 AM2/6/04
to
On Thu, 5 Feb 2004 22:09:04 +0000, Martin Brown
<|||newspam|||@nezumi.demon.co.uk> wrote:

>In message <vqb520p6d2gm6q6oc...@4ax.com>, jpc@?.?.invalid
>writes
>>On Thu, 05 Feb 2004 08:39:15 +0100, Hans-Georg Michna
>><hans-georgN...@michna.com> wrote:
>>
>>>Hans-Georg Michna <hans-georgN...@michna.com> wrote:
>>>
>>>Let me ask this way: Does the red filter (on the respective
>>>pixels) in a Bayer sensor let only red light pass? Or is it a
>>>pink filter that lets all colors pass, only more red than the
>>>others?
>>>
>>
>>If you want to see how much of each color the Bayer filters on your
>>camera passes or blocks do the following.
>>
>>
>>Download the tif image of visible spectrum you find at
>>http://home.earthlink.net/~skywise711/LasersOptics/Reference/reference.html.
>>It is a reasonablely intense image with wavelength markers
>>
>>It is also a long narrow image so I cropped and resampled it to print
>>it on a 4 by 6. You could also crop out the violet section beyond the
>>dark blue since that is only an artistic attempt at showing UV.
>>
>>Print it out and take an image under diffuse sunlight.
>
>I hope you are joking.

Not at all. Since the original poster had asked a legitimate question
and all he was getting in the post at the time was a continuation of
one of the sillier flame-wars I've seen I attempted to answer his main
question--do the transmittance of Bayer filters overlap. Which they
do,


>
>This will only show you the reflectivity of the ink system used by your
>printer when it tries to emulate the visual appearance of a solar
>spectrum. And it won't even do that particularly well.

As a rough and ready appoximation this method worked reasonably well


>
>If you want to characterise a camera's filters the simplest test at a
>consumer level is to photograph the reflection of a solar spectrum off a
>silver shovelware CD. You will likely be very surprised by the result.

And yes I hadn't thought of using a cd as a diffraction grating and
yes your method worked even better. With it I discovered the blue
filter does block the middle red completely instead of passing a few
percent like I stated earlier. It also passes the deep red and
infrared, something I'd already known from my infrared photography.

The big surprise what how much blue the green filer passes,
Both the red and blue filters look like they are single layer
interference filters. The green filter, on the other hand looks like
it is a multilayer since it transmittance peaks in the green than
drops about 30 % and stays pretty constant throughout the blue.


Why I don't know. The camera--an oly 3020Z --produces accurate colors
so it must be deliberate. If anyone knows something about how the
bayer calculations work jump in and educate us.


jpc

jpc

unread,
Feb 6, 2004, 12:57:41 PM2/6/04
to
On Fri, 06 Feb 2004 12:58:59 +0100, Hans-Georg Michna
<hans-georgN...@michna.com> wrote:


>pure curiosity. I'm wondering how much emphasis they put on high
>color resolution and precision.

My guess is that the bayer calculations won't work unless there is
some overlap
>

>
>Maybe one day we will have digital cameras with two kinds of
>interspersed sensors, one bunch for high quality, high light
>situations and a thinner grid of high sensitivity, lower quality
>sensors for low light situations. I would appreciate this and
>gladly dispense with the flash.

You have to trade resolution for S/N and dynamic range. And since it
is a square root function the resolution goes bad fast.


>
>I think there is at least one camera that already works that
>way, but I have no details.

Fugi came out with a fancy sensor last summer. No one in this news
group seemed that impressed. The one time I played with it I thought
it handled nicely but I never had a chance to take any images.

I also remember that at least one
>Sony camera has a special low light setting, but I don't know
>whether it actually has different kinds of sensors.
>

With the better Sony cameras and camcorders you can flip the ir filter
out of the light path. The sensor stays the same.


jpc

Bart van der Wolf

unread,
Feb 6, 2004, 5:53:38 PM2/6/04
to

"Hans-Georg Michna" <hans-georgN...@michna.com> wrote in message
news:43s320pbamhouj4ia...@4ax.com...
SNIP

> Let me ask this way: Does the red filter (on the respective
> pixels) in a Bayer sensor let only red light pass?

Mostly yes, but it (and the G and B filters) are not strict band pass
filters, there is some overlap in both G and B. The amount is determined by
the filters themselves, so only the manufacturer knows.

Bart


Bart van der Wolf

unread,
Feb 6, 2004, 5:59:12 PM2/6/04
to

"Guido Vollbeding" <gu...@jpegclub.org> wrote in message
news:40236764...@jpegclub.org...
> Lionel wrote:
SNIP

> The basic assumption of sampling theory is that every single sample
> sufficiently covers the range of susceptible values.

A 'monochrome' sensor does that, so your assumption needs some refining...

Bart


Gherry Bender

unread,
Feb 6, 2004, 9:56:17 PM2/6/04
to

> > Let me ask this way: Does the red filter (on the respective
> > pixels) in a Bayer sensor let only red light pass?
>
Another mis-conception seems to be here. Overlap is not a bad thing. It
makes for a more accurate reading as long as the overlapping portions are
not linearly related.

>


Guido Vollbeding

unread,
Feb 7, 2004, 9:32:07 AM2/7/04
to
Bart van der Wolf wrote:
>
> > The basic assumption of sampling theory is that every single sample
> > sufficiently covers the range of susceptible values.
>
> A 'monochrome' sensor does that, so your assumption needs some refining...

Yes, a 'monochrome' sensor does that, but in a Bayer sensor the color
masks
*prevent* that. Thus the Bayer system is a false sampling system.
It is to notice that the 'experts' speaking about sampling theory terms
like aliasing and Nyquist don't see that the Bayer system violates the
basic sampling assumptions and thus only produces artifacts.
Their theory makes them blind for the most obvious facts.

Regards
Guido

Lionel

unread,
Feb 7, 2004, 11:22:09 PM2/7/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

>Lionel wrote:


>>
>> >The art of Bayer sensing is a *major* violation of basic sampling
>> >assumptions
>>
>> Which ones? - Please quote the actual rules concerned, eg; Nyquist, etc.
>
>The basic assumption of sampling theory is that every single sample
>sufficiently covers the range of susceptible values.

That comment is so vague as to be nonsensical. What kind of sample? What
range? And what the hell is a 'susceptible value', exactly?

>The analog-digital conversion must result in quantized values which
>cover the susceptible range.
>This digital quantization of values is, beside resolution (digital
>quantization in space), one of the two basic sampling parameters.
>
>If the quantization range of single values is as limited as in the
>Bayer case, the other parameter, resolution in space of multiple
>values, does not make sense. That's why the 'MP' rating of Bayer
>sensors is wrong and misleading.

None of the above gibberish makes any sense. If you're going to try to
sound scientific, please try to use the correct terminology, rather than
making up your own.

Lionel

unread,
Feb 7, 2004, 11:45:49 PM2/7/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

>Bart van der Wolf wrote:

You still haven't explained this strange theory of yours in actual
scientific terms yet. So far, you haven't made any sense.

Guido Vollbeding

unread,
Feb 8, 2004, 9:01:36 AM2/8/04
to
Lionel wrote:
>
> None of the above gibberish makes any sense. If you're going to try to
> sound scientific, please try to use the correct terminology, rather than
> making up your own.

The basic fact to take care of is that the Bayer vendors use FAKE
Megapixel ratings! The Bayer system does produce Megasomethings,
but certainly not Megapixels!

All you naive users are fooled by the industry, and unfortunately
Foveon is therefore forced to fool you also with their upcoming
new products at PMA, so take care!
I'm currently trying as much as I can do to defuse their actual
policies which they are forced to pursue by OEM vendors due to
naive users, but I can only somewhat defuse, not prevent the
ongoing fakes and hoaxes, so take care!

Regards
Guido

Lionel

unread,
Feb 9, 2004, 10:05:38 AM2/9/04
to
Kibo informs me that Guido Vollbeding <gu...@jpegclub.org> stated that:

>Lionel wrote:


>>
>> None of the above gibberish makes any sense. If you're going to try to
>> sound scientific, please try to use the correct terminology, rather than
>> making up your own.
>
>The basic fact to take care of is that the Bayer vendors use FAKE
>Megapixel ratings! The Bayer system does produce Megasomethings,
>but certainly not Megapixels!

Well, whatever it is that they produce, there's more of them at higher
quality than your buddy George has been able to show us from his SD9, so
far.

>All you naive users are fooled by the industry,

'Naive'? I've designed image capture (laser scanning confocal
microscopy) systems myself. I don't claim to be an expert on image
sensors at the silicon level, but I'm hardly naive about the subject.

> and unfortunately
>Foveon is therefore forced to fool you also with their upcoming
>new products at PMA, so take care!

I'll take my chances on that one.

>I'm currently trying as much as I can do to defuse their actual
>policies which they are forced to pursue by OEM vendors due to
>naive users, but I can only somewhat defuse, not prevent the
>ongoing fakes and hoaxes, so take care!

Guido, if anyone could get idiots like George to stop making ridiculous
claims about the Foveon sensor, they'd be doing the company a major
service. All of us who're familiar with the technology know that the
Foveon shows a lot of promise, & certainly has the potential to be far
better than conventional sensors, but it's not yet there, & if people
keep on making claims for it that it can't yet live up to, it'll die in
the market before it's ever had a chance to reach that potential.

0 new messages