Digital P&S and color bit depth

Why is it arcane as a subject? The bigger those numbers the
more color there will be, right? Try to imagine the Canon D1
(if you know those specs) with 42 or 48 bit color depth, might
be quite awesome images from it (hypathetical). I think the Fuji
film D-SLRs put out a 42 bit raw file!

That's what I mean by the artistic thing, "a higher wealth of
color capability". The more the better for a D-P&S! Even then,
the better the B&W would be from Photoshop.

--
})))* Giant_Alex
cravdraa_at-yahoo_dot-com
not my site: http://www.e-sword.net/

I though thats what you were getting at. I agree, more bit depth is
needed, but it's more complicated than that.

For one thing, about 1/3 of the bit depth is useless, on the dark end
of the color. More bits help with that, giving you more upper end bits
to work with.

But the big sticking point is color interpolation, a "feature" of bayer
sensors. You do not get real-world, true pixel per pixel color from a
bayer sensor. Instead, you get an estimate, based on whatever
algorythm the camera maker uses, of the real color. The computer in
the camera fills in the blank spots with it's guess of what color was
really in that pixel. Unlike, say, slide film. Bayer sensors do not
have true color, the dif. color pixels are dispersed in a pattern.

Which is why I now shoot slide film now, and scan it in with a Nikon
Coolscan 5000. True color, because the scanner scans every color for
every pixel by stepping across the film. I can't stand digital color,
it's so blah. Nothing beats the black you get from slide film, either!
The trade off is a lack of sharpness, cause digital is sharp, got to
give it that. But I find a few sharpenings with a photo editor gets
pretty close. A full scan, from the Nikon scanner, yields a 131mb
file, from a 35mm slide. I find 5mb is ok for 4x6's. And you get real
bokeh back, too! So that's how I solved my artistic problem, I was not
happy with digital at all. In five years or so, that might change, new
sensors coming online.

On 27 Oct 2006 09:57:08 -0700, Xiaoding wrote:

Which is why I now shoot slide film now, and scan it in with a Nikon
Coolscan 5000. True color, because the scanner scans every color for
every pixel by stepping across the film. I can't stand digital color,
it's so blah. Nothing beats the black you get from slide film, either!
The trade off is a lack of sharpness, cause digital is sharp, got to
give it that. But I find a few sharpenings with a photo editor gets
pretty close. A full scan, from the Nikon scanner, yields a 131mb
file, from a 35mm slide. I find 5mb is ok for 4x6's. And you get real
bokeh back, too! So that's how I solved my artistic problem, I was not
happy with digital at all. In five years or so, that might change, new
sensors coming online.

Just a question about your mention of bokeh. Isn't that a
function of the lens? Meaning, whether used on a digital or film
camera, wouldn't the bokeh be the same?

On Fri, 27 Oct 2006 21:01:45 -0400, Ed Ruf (REPLY to E-MAIL IN
SIG!) wrote:

I believe you are confusing bit depth with dynamic range. Just what is a
film D-SLR? Seems to be an oxymoron.

Nope, nothing like that, if I might assume what AAvK meant by
that. Their film SLRs are Fujifilm SLRs, and their DSLRs are
Fujifilm DSLRs. If you look on Fuji product packaging, manuals and
on their website, the company is actually FUJIFILM. From my P&S
digital manual and brochure here's more than anyone but kinga needs:
(actually if anyone has all this, he's probably the one g)

Product Name: FUJIFILM DIGITAL CAMERA FinePix S5100 / FinePix S5500
Manufacture’s Name: Fuji Photo Film Co., Ltd.
Manufacture’s Address: 26-30, Nishiazabu 2-chome, Minato-ku,
Tokyo 106-8620, Japan
For more information on the full range of Fujifilm digital products,
please visit our Website: http://home.fujifilm.com

Xiaoding wrote:
Why is it arcane as a subject? The bigger those numbers the
more color there will be, right? Try to imagine the Canon D1
(if you know those specs) with 42 or 48 bit color depth, might
be quite awesome images from it (hypathetical). I think the Fuji
film D-SLRs put out a 42 bit raw file!

That's what I mean by the artistic thing, "a higher wealth of
color capability". The more the better for a D-P&S! Even then,
the better the B&W would be from Photoshop.

--
})))* Giant_Alex
cravdraa_at-yahoo_dot-com
not my site: http://www.e-sword.net/


I though thats what you were getting at. I agree, more bit depth is
needed, but it's more complicated than that.

For one thing, about 1/3 of the bit depth is useless, on the dark end
of the color.

I disagree with this statement. One needs to process the data
to display/print it, but it is not useless.

More bits help with that, giving you more upper end bits
to work with.

If you expose correctly. I still find it strange that people
with film experience continue to complain about dynamic range
of digital when digital has a much higher dynamic range than
film, print or slide. Learn to use your light meter correctly
for the medium. Just like print film must be exposed
differently than slide film, digital is different from either.

But the big sticking point is color interpolation, a "feature" of bayer
sensors. You do not get real-world, true pixel per pixel color from a
bayer sensor.

And neither does your eye. You eye is like a Bayer sensor too!

Then, what is true color? If you read how the eye works, you would
find it is actually impossible for film or digital to produce
accurate colors we see, because our eye is non-linear and subtracts
color from different receptors. What is amazing is that color
works as well as it does. There are whole books written on this
subject.

Instead, you get an estimate, based on whatever
algorythm the camera maker uses, of the real color. The computer in
the camera fills in the blank spots with it's guess of what color was
really in that pixel.

Digital cameras also use a blur filter, which acts to spread the
light from one spot in the image over the RGB pixels. This averaging
helps to equalize the discrete colored pixels to give better
color reproduction.

Unlike, say, slide film. Bayer sensors do not
have true color, the dif. color pixels are dispersed in a pattern.

And film has true color? Light entering the film emulsion is
subject to scattering and absorption by the dye clouds. If you actually
looked up tests of color accuracy of film and digital, you would
find that the digital cameras are consistently better at color accuracy
than any film.

Which is why I now shoot slide film now, and scan it in with a Nikon
Coolscan 5000. True color, because the scanner scans every color for
every pixel by stepping across the film.

You think slide film is true color?

I can't stand digital color,
it's so blah.

Ah, now we see! You don't actually want "true" color. You want
vivid color. That's not true color.

Nothing beats the black you get from slide film, either!

Sorry, but your wrong again. You can dig deeper into shadows and
dark areas with digital, and produce much smoother blacks than any
slide film. Learn to process your digital images. Film has a toe to
it characteristic curve, digital camera images do not. Add a toe in
your photo editor with the curves tool.

The trade off is a lack of sharpness, cause digital is sharp, got to
give it that. But I find a few sharpenings with a photo editor gets
pretty close.

Most photo editors do not have any actual sharpening tools (photoshop
does not). Tools like unsharp mask do not sharpen, they change
accutance.

A full scan, from the Nikon scanner, yields a 131mb
file, from a 35mm slide. I find 5mb is ok for 4x6's. And you get real
bokeh back, too! So that's how I solved my artistic problem, I was not
happy with digital at all. In five years or so, that might change, new
sensors coming online.

This might be a statement from 5 years ago. It hasn't been true for years.
One can scan a piece of film at any megapixel you want. I've done
11,000 dpi drum scans of 35mm Velvia, and guess what? There is no
more information than at 6,000 ppi. Consensus by many photographers
is that around 8 megapixels, DSLRs equal or beat 35mm ISO 100 slide film
in terms of resolution.

Some references relevant to the OP and this response:
http://www.clarkvision.com/imagedeta....summary1.html

Dynamic Range and Transfer Functions of Digital Images
and Comparison to Film
http://www.clarkvision.com/imagedetail/dynamicrange2

Digital Cameras: Does Pixel Size Matter?
Factors in Choosing a Digital Camera
http://www.clarkvision.com/imagedeta...el.size.matter

Roger
Photos, digital info at: http://www.clarkvision.com

wrote:

David J Taylor wrote:

wrote:


The Lecia DMR produces 16 bit/channel RAW files. How much of those 16
bits is usable, I don't know.

Thanks for that. I couldn't find that camera listed at D P Review. Do
you happen to know how many bits the camera's ADC uses?

David


You can read about the DMR he
http://www.leica-camera.us/photograp...gital-modul-r/

The DMR uses a Kodak KAF-10010 CCD image sensor. From the data sheet,
the sensor has a dynamic range of 67 dB. If I remember my signal
processing facts, that corresponds to a hair over 11 bits. The data
sheet is he
http://www.kodak.com/ezpres/business...10LongSpec.pdf

BTW, the Lecia M8 uses a newer KAF-10050 sensor. It has a dynamic
range of 76 dB, almost 13 bits. The M8 is on dpreview:
http://www.dpreview.com/reviews/spec...a/leica_m8.asp

Interesting.

With 6.8 micron pixels, and 40,000 electron full well, and assuming
ISO 100 is set to maximize the full well capacity, that sets the
12-bit unity gain at just under ISO 1,000. That would plot
just to the lower right of the Canon 350D on Figure 6 at:
http://www.clarkvision.com/imagedeta...el.size.matter

I searched for a data sheet on the KAF-10050 sensor, but could
not find one. The data sheet for the KAF-10010 is great and more
complete than older Kodak data sheets. The off-axis response
could have helped in a long thread on purple fringing in the NG
a few months ago.

6.8 microns and 40,000 electron full well is about average for chips
that size, but the 17 electron read noise of the KAF-10010 is high,
especially compared to Canon's CMOS chips which run under 4 electrons.
Presumably the KAF-10050 has improved the read noise, but the full well
is probably about the same as the KAF-10010 at about 40,000.

Roger

I though thats what you were getting at. I agree, more bit depth is
needed, but it's more complicated than that.

For one thing, about 1/3 of the bit depth is useless, on the dark end
of the color. More bits help with that, giving you more upper end bits
to work with.

But the big sticking point is color interpolation, a "feature" of bayer
sensors. You do not get real-world, true pixel per pixel color from a
bayer sensor. Instead, you get an estimate, based on whatever
algorythm the camera maker uses, of the real color. The computer in
the camera fills in the blank spots with it's guess of what color was
really in that pixel. Unlike, say, slide film. Bayer sensors do not
have true color, the dif. color pixels are dispersed in a pattern.

Which is why I now shoot slide film now, and scan it in with a Nikon
Coolscan 5000. True color, because the scanner scans every color for
every pixel by stepping across the film. I can't stand digital color,
it's so blah. Nothing beats the black you get from slide film, either!
The trade off is a lack of sharpness, cause digital is sharp, got to
give it that. But I find a few sharpenings with a photo editor gets
pretty close. A full scan, from the Nikon scanner, yields a 131mb
file, from a 35mm slide. I find 5mb is ok for 4x6's. And you get real
bokeh back, too! So that's how I solved my artistic problem, I was not
happy with digital at all. In five years or so, that might change, new
sensors coming online.

You know, I fully agree with you from an artistic standpoint.
That other tech_maniac_guy, good grief.

I accept the idea of scanning positive and negative (and I do it)
more than a dij-cam because these colors are pure by develop-
ement in a natural, flowing chemical processes and can be
scanned in 48 bit color.

But you've got to realize, I have a Umax powerlook III and
I didn't know but I suspected it had a "bar of CCD sensors"
going across the 8-1/2" width to scan... I took it apart to clean
it internally - nope, it's got a small square CCD behind a round
1 inch O.D. lens, the lamp reflects light off the subject - the
image is concentrated through a complex of bowed silver
surfaced mirrors and into that lens to the CCD.

But it is a "CCD" which means Bayer, just like in your Nikon
5000, though yours doesn't have the mirrors. Why it is better
than a dij-cam, you can get your image as a more pure form of
color, and 48 bit color depth. If I had $1900.00 I would buy
that big current one (9000?) no sweat, I'd like to get the V ED
just for my current collection of 35 negs and poz's, and keep
on shooting 35!

So, I'm not going to defend dij-cams compared to fine scanning,
no way. But that $550 - $1900 Nikon scanner is too expensive
and a dij-cam with 12 bits per RG and B is cheaper and far more
convenient to simply get the image onto my HD, you know. My
point is, as_good _a_quality_as_possible, economically.

--
})))* Giant_Alex
cravdraa_at-yahoo_dot-com
not my site: http://www.e-sword.net/

"AAvK" writes:

I accept the idea of scanning positive and negative (and I do it)
more than a dij-cam because these colors are pure by develop-
ement in a natural, flowing chemical processes and can be
scanned in 48 bit color.

So chemical processes are more "natural" than electronic ones? The way
that CCDs convert photons into electrons is just as natural as the way
that photons chemically change silver bromide crystals so they can be
developed.

But you've got to realize, I have a Umax powerlook III and
I didn't know but I suspected it had a "bar of CCD sensors"
going across the 8-1/2" width to scan...

There *are* scanners like that, e.g. the Canon LiDE scanners.
But that has its own advantages and disadvantages.

I took it apart to clean
it internally - nope, it's got a small square CCD behind a round
1 inch O.D. lens, the lamp reflects light off the subject - the
image is concentrated through a complex of bowed silver
surfaced mirrors and into that lens to the CCD.

If you look carefully, you'll almost certainly find that all the mirrors
are flat front-surface ones. They only act to fold the light path into
a small space. All the focusing is done by the lens.

But it is a "CCD" which means Bayer, just like in your Nikon
5000, though yours doesn't have the mirrors.

Here you're dead wrong. CCD just refers to the light-measuring
technology, and there are CCDs with and without Bayer filters.
In particular, the CCDs used in flatbed and film scanners are
"trilinear" arrays, which have 3 rows of sensing locations. Colour
comes from a red filter in front of one row, green in front of another,
and blue in front of the third row. (There's sometimes a 4th row I
won't get into here). So every time the scanner head moves, it captures
three different lines on the page in three different colours. After
some processing to line up the three images, you end up measuring all 3
colours at every pixel.

Why it is better
than a dij-cam, you can get your image as a more pure form of
color, and 48 bit color depth.

48 bit is just the output bit depth. Plenty of digital cameras can
output 48-bit image files when shooting in RAW mode. But the A/D
converter resolution is less, probably about 12 bits in both cases.

Dave

On Sun, 29 Oct 2006 21:43:42 +0000 (UTC), (Dave
Martindale) wrote:

"AAvK" writes:

But it is a "CCD" which means Bayer, just like in your Nikon
5000, though yours doesn't have the mirrors.

Here you're dead wrong. CCD just refers to the light-measuring
technology, and there are CCDs with and without Bayer filters.
In particular, the CCDs used in flatbed and film scanners are
"trilinear" arrays, which have 3 rows of sensing locations. Colour
comes from a red filter in front of one row, green in front of another,
and blue in front of the third row. (There's sometimes a 4th row I
won't get into here). So every time the scanner head moves, it captures
three different lines on the page in three different colours. After
some processing to line up the three images, you end up measuring all 3
colours at every pixel.


And just to be clear, if it's a Nikon film scanner like
the LS-8000/9000, it's a *monochrome* CCD with
no filters at all over it. Instead, the Nikons use three
sets of LEDs -- red, green and blue, and alternate
these rapidly -- just as in a Canon LIDE flatbed scanner,
though the Canon uses a CIS sensor, rather than CCD.

In the LS8000/9000 there is exacly one mirror and
it is a small perfectly flat front-surface mirror.

The lens in the LS8000/9000 is quite a beauty.
Probably a lot more sophisticated than even my
favorite Canon L zoom lens.

Film scanners *do NOT* use Bayer sensors.


rafe b
www.terrapinphoto.com

ASAAR wrote:
On 27 Oct 2006 09:57:08 -0700, Xiaoding wrote:

Which is why I now shoot slide film now, and scan it in with a Nikon
Coolscan 5000. True color, because the scanner scans every color for
every pixel by stepping across the film. I can't stand digital color,
it's so blah. Nothing beats the black you get from slide film, either!
The trade off is a lack of sharpness, cause digital is sharp, got to
give it that. But I find a few sharpenings with a photo editor gets
pretty close. A full scan, from the Nikon scanner, yields a 131mb
file, from a 35mm slide. I find 5mb is ok for 4x6's. And you get real
bokeh back, too! So that's how I solved my artistic problem, I was not
happy with digital at all. In five years or so, that might change, new
sensors coming online.

Just a question about your mention of bokeh. Isn't that a
function of the lens? Meaning, whether used on a digital or film
camera, wouldn't the bokeh be the same?

Well, yes, in theory. But your average P&S has a small sensor, which
means less bokeh. Full size sensors would have the same, but then you
got to spend a lot more money, and you still got the color problems.
So I went for the scanner, since I already had film gear.

Roger N. Clark (change username to rnclark) wrote:
I searched for a data sheet on the KAF-10050 sensor, but could
not find one.

It's here.
http://www.kodak.com/ezpres/business...00LongSpec.pdf