Technical inaccuracy in Foveon press release?

[Ted B.47613]

Perhaps your glasses need cleaning?

I can easily distinguish an in-camera jpeg image (D30) from a raw-to-tiff image. The jpeg artifacts are obvious. Shadow noise increases as well, as does flat-area noise.

A Canon D60 will store a 6MP fine JPEG at 2.5MB per image(which is

just about indistinguishable from RAW). Or about 7.5MB for the raw
format.

Peter

--
'Think Outside The Box.........Once you're in the box, it's too late.'

My Galleries: http://home.attbi.com/~keylargographics/

 

[Peter G]

Bob,

My point about the captured images is that people sometimes people don't realize how good the quality of JPEG is.

As for the format the camera can store in, I am going by the press release, info of Sigmas site, and Phils summary of the SD9 which says:

Uncompressed Format: Raw.
Compressed Format: NONE!

http://www.dpreview.com/reviews/compare_post.asp?cameras=sigma_sd9&method=sidebyside

Of course you can turn it into a JPG or whatever you want once it is on your computer. That is not the issue. The issues is the relatively high price of flash memory for storing 9mb/image.

I mean you are only going to get about 13 images on a 128MB flash card. Even if you have a digital wallet you will need 256MB flash card minimum just so you don't get constantly interuppted to empty the card.

This is a very big issue IMO and I hope it is something they fix before release.

I suppose it is a non issue if you only want to take pictures tethered to a computer. I know I want to take pictures outdoors, so storage is a big issue.

Peter

Of course with a Foveon-sensored camera the Raw format will occupy
more room than a JPEG because there are up to three words a pixel.

Size issues apart, raw format will still have advantages in better
post-processing control.

--
Derek

 

[Peter G]

I read both Canon forums regularly. I have seen this comparison done several times and there is virtually no jpg artifacts on Canon Super fine jpegs. General concensus is that RAW's main advantage is the post production cabability.

There is certainly more bit depth in a D30 RAW file, but you are limited to 8 bits when viewing anyway. I am big proponent of RAW capture. But also think they need to include a format with a size a little smaller than 9MB. Some of us can't afford 1GB flash cards.

Send me a .BMP file with a section from one of your TIFF's where you expect obvious JPG artifiacts. I will convert it to a high quality JPG and back to .BMP and then compare them.

If you want to suggest the in-camera JPG is the problem that is not the issue. JPG can be very high quality and space efficient at its highest setting, artifacts take a 400% zoom and time on your hands to find, so they are negligible IMO.

Peter

I can easily distinguish an in-camera jpeg image (D30) from a
raw-to-tiff image. The jpeg artifacts are obvious. Shadow noise
increases as well, as does flat-area noise.

A Canon D60 will store a 6MP fine JPEG at 2.5MB per image(which is

just about indistinguishable from RAW). Or about 7.5MB for the raw
format.

Peter

--
'Think Outside The Box.........Once you're in the box, it's too late.'

My Galleries: http://home.attbi.com/~keylargographics/

 

[David Barkin]

Hi Bob

I wasn't posing the question to argue with you. I absolutely take your word for it. I was left handedly critiqueing your writing and formatting style. And also asking a real question so that your answer would clarify what you said.

Toss in a few line breaks, probably would help.

Since I purchased a D1x and find it an excellent camera, your post is in the class of "Ok sucker, here's the bad news on your fancy, shamancy camera! I don't believe in shooting the messinger, although feeding you to my dog is a possibility.

One other thing I did when I bought the Nikon was to buy all Sigma lenses. These of course have Nikon mounts. Can I use these if I bought a Sigma now?

Dave

repeat, The loss of resolution caused by the Bayer pattern coupled
with the errors introduced by the interpolation step in things like
edges, makes the 3.4 mp Foveon the equal of a 6mp Bayer sensor,
only with better color, under most circumstances. Whether or not
Sigma can put that power to use is yet to be seen.

Hi Bob

I lost it in your long thread. But I have one question. All things
be equal and I shoot a picture of my dog at 100 feet with a 6 MEG
CCD sensor and a 3 Meg foveon sensor using a "normal" lens. I load
the image into Photoshop and then zoom in on the dog.

Assuming I can see the grain of the dogs hair with the 6 Meg sensor
(and I can) could I also see it on the foveon sensor? I know that
with a three Meg ccd I cannot.

Dave

 

[Peter G]

From all the information I looked at, I believe the answer to your question is yes. I expect a 3.3MP X3 image to equal or surpass a 6MP mosaic image in all respects.

I was heavily involved in most of the early X3 threads, usually as a devils advocate, quelling what I thought was an over-reacation to this technology.

In terms of absolute detail. Mosaic tends to top out at 75% of the detail warranted by their number of pixels in a linear direction. I measured it on several res charts during the initial arguments.

OTOH X3 cameras are capable of resolving all 100% of the detail for their pixel dimensions.

To map that to pixel equivalences you need both dimensions. .75x .75 = .56. So 0.56 is the empirical factor.

A 6.0 MP mosaic camera approximates ( 6.x .56) = 3.36MP of X3 type detail.

Not only that but X3 looks a lot clearer when aproaching those limits. So I have little doubt that with an equivalent optical path the X3 tech 3.3 will be more than competitive with the 6.0MP mosaic cameras.

But I still think there is a ways to go, before X3 will be in my hands. We buy camears after all, not sensors.

There have been great cameras and terrible cameras using exactly the same sensors. That won't change with X3. There have been cameras that take good pictures but have terrible interfaces and usability. That won't change with X3.

Unfortunately I like Canons features right now, and I think they will be late to go X3 since they have their own CMOS investment. Maybe in three or 4 years I can get a Canon D400 with 6MP X3 for $1500.

Peter

Hi Bob

I lost it in your long thread. But I have one question. All things
be equal and I shoot a picture of my dog at 100 feet with a 6 MEG
CCD sensor and a 3 Meg foveon sensor using a "normal" lens. I load
the image into Photoshop and then zoom in on the dog.

Assuming I can see the grain of the dogs hair with the 6 Meg sensor
(and I can) could I also see it on the foveon sensor? I know that
with a three Meg ccd I cannot.

Dave

 

[Jason Busch]

Kevin,

It think the Foveon chip is a great new technology, but I am wondering why the limited ISO settings on the SD9, the first camera to have a Foveon chip in it. Not to mention, if the SD9 is going to compete with the D100 and D60, you would think they would include more ISO settings. Normally, a manufacturer would include these settings, if they were possible, because it brings them to the same level as their competitors. So, this has me puzzled. I am assuming that the Foveon Chip has a very limited ISO, because of the decision of Sigma to put a wider range of ISO settings on their first Foveon Technology camera. I will think more of the SD9 if they change this spec, but I have my doubts. My theory is this. The Foveon Chip is extremely dependent on light. More so than a CCD or CMOS chip. Because of this, lower light photos actually increase the noise by 3x the normal rate. There is an extreme drop-off curve of ISO quality, when compared to CCD and CMOS chips.

Now, this is just a wild theory, but like I said, until Sigma changes the spec on that camera, I can't think of any reason why they would not include higher ISO settings.

--
Forum:
http://pub103.ezboard.com/bthedigitaldinguscommunity

Websites:
http://e10club.topcities.com/
http://d100.topcities.com/
--

'I do just about everything in my CCDs...'

 

[Zoltan Csuka]

Everybody has been wondering about the ISO limitations for months now. We will see.

Kevin,

It think the Foveon chip is a great new technology, but I am
wondering why the limited ISO settings on the SD9, the first camera
to have a Foveon chip in it. Not to mention, if the SD9 is going to
compete with the D100 and D60, you would think they would include
more ISO settings. Normally, a manufacturer would include these
settings, if they were possible, because it brings them to the same
level as their competitors. So, this has me puzzled. I am assuming
that the Foveon Chip has a very limited ISO, because of the
decision of Sigma to put a wider range of ISO settings on their
first Foveon Technology camera. I will think more of the SD9 if
they change this spec, but I have my doubts. My theory is this. The
Foveon Chip is extremely dependent on light. More so than a CCD or
CMOS chip. Because of this, lower light photos actually increase
the noise by 3x the normal rate. There is an extreme drop-off curve
of ISO quality, when compared to CCD and CMOS chips.

Now, this is just a wild theory, but like I said, until Sigma
changes the spec on that camera, I can't think of any reason why
they would not include higher ISO settings.

--
Forum:
http://pub103.ezboard.com/bthedigitaldinguscommunity

Websites:
http://e10club.topcities.com/
http://d100.topcities.com/
--

'I do just about everything in my CCDs...'

--
Best wishes,
Zoli

 

[David Barkin]

Well here's another question. Scanners use the same ccd technology that most camera's use. Is anyone going to apply it to a scanner?

Dave

 

[Kevin Chu]

Hey, good theory. I don't remember all the terminology, but if one sensor receives more light than an adjacent sensor, the difference shows up in the image. If the sensors aren't receiving much light (so it's probably dark and you're using a higher sensitivity), then any difference between the amounts received by adjacent sensors becomes more significant and has a greater impact on the image. For sensors that sense three times at a single location, the magnitude of the problem may triple.

But.... For Foveon sensors that sense three times at the same location, it's three different colours that are sensed. If the difference in the amounts received by adjacent sensors is not in all three colours.... For mosaic sensors, it's almost the same problem, but the sensors for the three colours are in three separate places, which may or may not have similar effects.

My mind boggles. I think I'll just keep snapping away with my CP995 and wait and see. The D100 sure is nice though.

Kevin

Kevin,

It think the Foveon chip is a great new technology, but I am
wondering why the limited ISO settings on the SD9, the first camera
to have a Foveon chip in it. Not to mention, if the SD9 is going to
compete with the D100 and D60, you would think they would include
more ISO settings. Normally, a manufacturer would include these
settings, if they were possible, because it brings them to the same
level as their competitors. So, this has me puzzled. I am assuming
that the Foveon Chip has a very limited ISO, because of the
decision of Sigma to put a wider range of ISO settings on their
first Foveon Technology camera. I will think more of the SD9 if
they change this spec, but I have my doubts. My theory is this. The
Foveon Chip is extremely dependent on light. More so than a CCD or
CMOS chip. Because of this, lower light photos actually increase
the noise by 3x the normal rate. There is an extreme drop-off curve
of ISO quality, when compared to CCD and CMOS chips.

Now, this is just a wild theory, but like I said, until Sigma
changes the spec on that camera, I can't think of any reason why
they would not include higher ISO settings.

--
Forum:
http://pub103.ezboard.com/bthedigitaldinguscommunity

Websites:
http://e10club.topcities.com/
http://d100.topcities.com/
--

'I do just about everything in my CCDs...'

 

[Rinus Borgsteede]

Low ISO is natural for a multi layered CCD. Just like in film, the CCD must employ filters between the layers to get only the correct color to the three apropriate layers. If the three layers were to be sensitive to different frequencies of the spectrum and so circumvent filters, the light would still have to pass trough a layer or two to get to its own color sensitive pixel. The sensitivity would have to be adequate for the lowest layer so as not to produce (too much) noise. The other two layers would have to be less sensitive. The top layer would receive the most light and I am sure it would haver to be attenuated so as not to be too strong in comparisson to the others. I am assuming here that the Foveon works in this fashion. I have seen the simple cross section of the chip design but no further explanation has been given as to how they want to accomplish the combined capture of light.
Rinus

 

[Peter G]

Scanners don't use a mosaic. They scan 3 colours at each pixel.

Well here's another question. Scanners use the same ccd technology
that most camera's use. Is anyone going to apply it to a scanner?

Dave

 

[Bob Williams]

David, it has been a long time since anyone paid me for writing.
I know that I tend to use long compound complex sentences.

No, I am not poking a stick at the people that bought the D30,
D60, or one of the more expensive cameras. Moore's law is at
work. In the beginning, only the CCD could be used, as CMOS
was too noisy. About 18 motnhs ago, Canon changed that by
developing a sensor with noise reduction circuitry on top of the
chip. The 18 month cycle has come around again. This time
the big step forward is the three level chip, built on a CMOS
process. This is the first actual usage of the tech, and there are
some rough edges. The ISO range is not big enough for some.
Being limited to the Sigma mount is not enough for others. Some
have said and written in published articles, that no pro would
ever show up at a shoot with a camera that says Sigma on the
front. No one is saying that there is no place for the other cameras
in this generation. The Foveon/Sigma just does what I want
my camera to do.

Yes, it has been reported that Sigma can and will change the mounts
on the lenses. I can not confirm this, as no one has actually tested
it that I know of.

Peter, you are correct in that there is very little difference between
RAW and high quality jpg's on most of todays cameras, and that the
primary advantage other than the smoother color, is the ability to
adjust the shoot settings after capture. This is because the Bayer
interpolation already introduces the errors into the files when you
are using a Bayer sensor. RAW does not stop the Bayer process.
The Foveon does not suffer from this limitation. The Bayer
process artifacts are not already present. You can actually use
and gain the benefit from the much greater amount of information
present in the RAW files. You are also correct in that for web
usage, a lot of that information is useless. If the web was my
main target, other than the obvious lens limits, the G1 I already
own is very good. For the lowest common usage these files will
be used for, I guess JPG would be fine. I print at least an 8x10
of all of my best shots. I often print multiple copies. Serious
reproduction of the images is where the full information is needed.
I will conceed that you and several other seem to be very worried
about the file size. Maybe they should add a jpg mode for those
people that would use it. I am not one of them. 2000$ camera,
900$ lens, and I am not worried about having to purchase a
300$ storage device to use it.

The resolution of the present D60 files being displayed is impressive.
The smoothness of crops blow way up is also. I am not knocking
the D60. I want a Foveon based camera, because it matches
my usage perfectly. If jpg, high ISO, Canon, Nikon, or even
Tamron lenses all are part of your needs, then this camera will
not be for you. Other than that, we do not know yet!

Hi Bob

 

[Thomas Sapiano]

I'm not 100% sure, however I believe that the Foveon chips are simply relying on the varrying absorbtion of light at different frequencies as the penetrate the surface of the solid state electronics - there are no filters built into the structure, the top level absorbs one band of freuqencies, while higher energy photons just tunnel right through, and are absorbed by the following sensors. The only photons that would be blocked by the higher layers would be those that are absorbed by them.

Low ISO is natural for a multi layered CCD. Just like in film, the
CCD must employ filters between the layers to get only the correct
color to the three apropriate layers. If the three layers were to
be sensitive to different frequencies of the spectrum and so
circumvent filters, the light would still have to pass trough a
layer or two to get to its own color sensitive pixel. The
sensitivity would have to be adequate for the lowest layer so as
not to produce (too much) noise. The other two layers would have to
be less sensitive. The top layer would receive the most light and I
am sure it would haver to be attenuated so as not to be too strong
in comparisson to the others. I am assuming here that the Foveon
works in this fashion. I have seen the simple cross section of the
chip design but no further explanation has been given as to how
they want to accomplish the combined capture of light.
Rinus

 

[GaryW]

Peter G wrote:
....

In terms of absolute detail. Mosaic tends to top out at 75% of the
detail warranted by their number of pixels in a linear direction. I
measured it on several res charts during the initial arguments.

OTOH X3 cameras are capable of resolving all 100% of the detail for
their pixel dimensions.

To map that to pixel equivalences you need both dimensions. .75x
.75 = .56. So 0.56 is the empirical factor.

A 6.0 MP mosaic camera approximates ( 6.x .56) = 3.36MP of X3 type
detail.

....

With all of this talk about .75 and .56, which determines the effective resolution of most cameras, can someone explain how black & white mode might change those numbers? Would you obtain the full resolution, and effectively use 100% of the pixels, not .75? It might give me enough excuse to try B&W some more.....

And if B&W is 1.0 and color is .75, what if you just lower the contrast? Does that make it .85? But you'd probably mess it all up again if you tried adjusting the contrast in the paint program, as you just don't have the data to recover.

--
Gary W.
Nikon 880

 

[David Rosenthal]

First, their statement is reasonable and correct--and hints at the reason behind it:

The answer lies in the fact that JPG compression isn't smart enough to "undo" the Bayer Interpolation .

Consider a 6mp regular camera. It measures 6 million total values but generates 18 million: 6 million reds, 6 million greens, and 6 million blues. This is from the Bayer interpolation process. When this image is JPG compressed the JPG works on the 18 million values. The compression isn't "smart enough" to see that the detail generated by the Bayer process isn't real and needn't be preserved. A larger than needed file results.

Now consider a 6mp ("2mp" equilivant) x3 camera. It also measures 6 million total values but only generates 6 million. The JPG works on these 6 million values and treats them all equally. The compression ratio will be slightly worse than images coming out of a "2mp" camera but not enough worse to compenste for the factor of 3 in generated data.

Hope this helps--this is a complicated subject. In short: the JPG process, from an entropy standpoint, is not totally efficient in its compression when applied after an interpolation process.

Dave

In http://www.foveon.com/press_X3_business.html , Foveon claims that:

"By capturing three colors at every pixel instead of just one,
Foveon X3 based cameras have measured information for all three
colors. This results in high quality photographs with fewer numbers
of pixels because the pixels are based on real measurements. Fewer
pixels results in smaller file sizes allowing digital camera users
to send a higher quality photograph more quickly through e-mail. It
also means that more digital photos can be stored on a digital
camera storage card."

Now can someone help me wrap my head around the concept of smaller
file sizes through the use of the Foveon sensor?

The way I understand it, for a given pixel in an image, you have
(three times the colour depth) bits per pixel. An image captured
by the Foveon sensor would have each of the three colour values per
pixel. An image captured with a standard CCD would have
information for each pixel composed of values measured by separate
R, G, and B sensors. So if all of the sensors were the same size,
the standard CCD would have to be three times as dense, but for
each pixel, there is still the same number of bits for each of R,
G, and B.

Kevin

 

[JerryKS]

Here is a URL that explains very well how scanners work.
http://www.sony.co.jp/~semicon/english/img/sony01/a6802815.pdf

Well here's another question. Scanners use the same ccd technology
that most camera's use. Is anyone going to apply it to a scanner?

Dave

 

[David Rosenthal]

Oh yea, and the e-mail thing is wrong--after you res-down any of these images a decent amount you are dealing with the same thing.

Fewer
pixels results in smaller file sizes allowing digital camera users
to send a higher quality photograph more quickly through e-mail.

 

[David Barkin]

Hi Bob

Actually of course your writing isn't that bad. I lost it because of the lack of line breaks. Reading off a monitor is quite different from reading off a page. As part of your post point out, there's more to a camera then just a sensor. I can live with the Nikon - It's a great camera and no doubt in two three years this new technology will be much more common. No matter how good or bad the Sigma is I doubt if it will be as "tough" as the Nikon.

Of course this did start a tremendous dispute with the dog who always gets punished for my errors! He won.

Dave

David, it has been a long time since anyone paid me for writing.
I know that I tend to use long compound complex sentences.

 

[Peter G]

Peter, you are correct in that there is very little difference between
RAW and high quality jpg's on most of todays cameras, and that the
primary advantage other than the smoother color, is the ability to
adjust the shoot settings after capture. This is because the Bayer
interpolation already introduces the errors into the files when you
are using a Bayer sensor. RAW does not stop the Bayer process.
The Foveon does not suffer from this limitation. The Bayer
process artifacts are not already present. You can actually use
and gain the benefit from the much greater amount of information
present in the RAW files. You are also correct in that for web
usage, a lot of that information is useless. I print at least an 8x10
of all of my best shots. I often print multiple copies. Serious

Bob, I am just wondering what the Case is for Not including JPEG. If the format is near indistinguishable when examining the image electronically, it will be likewise when printed. Minus the obvious greater image processing options that exist with RAW.

I have many times argued that RAW should be an option on all camears, due to the post capture flexibility. But I would never propose doing away with JPG.

This just strikes me as a serious gaff. This camera is out of my price range anyway so it is a theoretical discussion. I am curious about the buisness case for not having a useful compressed format for use when storage space is a limitation. I don't know anyone who owns a digicam who doesn't want more storage space.

When I sold my Nikon 950 the guy trying it was pleased that he could lower the resolution to 1024x768 and increase the compressiong to quintuple the amount of captured images. I was kind of stunned since, I never had never taken a picture at less than full resolution and best jpg.

I'll be curious to see if the first consumer X3 cameras have JPG.

Peter

 

[GaryW]

David Rosenthal wrote:
...

Consider a 6mp regular camera. It measures 6 million total values
but generates 18 million: 6 million reds, 6 million greens, and 6
million blues. This is from the Bayer interpolation process. When
this image is JPG compressed the JPG works on the 18 million
values.

....

Now consider a 6mp ("2mp" equilivant) x3 camera. It also measures
6 million total values but only generates 6 million. The JPG works
on these 6 million values and treats them all equally. The
compression ratio will be slightly worse than images coming out of
a "2mp" camera but not enough worse to compenste for the factor of
3 in generated data.

...

That is an interesting way to look at it. Although, based on the ".56" figure that others posted above, I would stick to the theory of there being more of a 2 times difference, not 3 times. So, your 2mp x3 camera has the filesize of a 2mp nomal camera, but the quality/detail of a 4mp, not 6. (See other messages for the derivation of the .56 figure.)

So, if I'm getting an inferior, interpolated picture from my camera, can I resize it to 75% (height & width), with no loss of real information? Why don't they do that in-camera? So they can quote more impressive pixel counts?

(You might be wondering why I would want to bother reducing pixels. Even if they are "extra", as long as you have them,you might as well keep them, right? Maybe, but then if you're going to manufacture information, you could use a different technique, such as Genuine Fractals. Just a thought....)
--
Gary W.
Nikon 880