Still confused about RAW & TIF

Floyd L. Davidson wrote:

If you are totally unaware of how digital imaging systems
function you should *not* be pontificating about the
differences. Saying there is no analog level is beyond simple
ignorance, it is an indication that you have no idea how a
digital imaging system works.


And EVERYTHING that I have read [and there is not reason to believe otherwise
as it just doesn't make sense] is that White Balance information is stored as
encrypted metadata in the RAW file. It is NOT applied until conversion time.

Pontificate all you want.

"As you can see from the diagram above the D2X reads data
from the sensor using four channels. To keep color
information as pure as possible this is done by grouping
pixels together by color; R, Gr, B, Gb. This four channel
readout is then fed to the analogue processing system / A/D
converter as separate channels and is processed as such. This
allows the camera to achieve its five frames per second
continuous shooting rate.


I didn't argue anything about 4-channels. I argued that White Balance
information is NOT applied in camera to a RAW file. A/D conversion is just
counting photons [and noise].


"The diagram above clearly demonstrates the D2X's four
channel readout and independent color processing and A/D
conversion which is carried out. Although Nikon aren't
revealing a huge amount of detail about what goes on at the
'pre-conditioning' stage (just before A/D) the idea is that
data from the individual channels can be amplified
specifically to apply some amount of WB correction and to
maximize color gradations before conversion to a digital
value in the A/D converter.

The diagram is poor, and does not indicate what the text describes.


In fact, that text specifically says that they do not know what happens in
"pre-conditioning" and simply surmise that some amount of White Balance
information is applied. Clearly ... that remains a mystery. The commonly
heald believe, from what I have read is that it is not.

A 16 bit TIFF formatted file is easily able to contain all of
the data available in a 12 bit NEF file.

I indicated that they would be similarily sized files for the same bit-depth.
They would.

They won't. A D2x produces a 4312x2868 pixel image, which
contains 12.37 MP. The file size if the non-compressed NEF mode
is selected will be roughly 20Mb per image. Converted to 16 bit
TIFF format that will produce a 37Mb file. If compressed it
will be slightly smaller 30-35Mb.)


I mis wrote there, as I meant to say analgous bit-depths. Clearly, a 12-bit
RAW is NOT the same thing as a 12-bit TIFF.

--
Thomas T. Veldhouse
Key Fingerprint: D281 77A5 63EE 82C5 5E68 00E4 7868 0ADC 4EFB 39F0

"Thomas T. Veldhouse" wrote:
Floyd L. Davidson wrote:

If you are totally unaware of how digital imaging systems
function you should *not* be pontificating about the
differences. Saying there is no analog level is beyond simple
ignorance, it is an indication that you have no idea how a
digital imaging system works.

And EVERYTHING that I have read [and there is not reason to believe otherwise

You have been provided with a clue for finding sufficient
reading material, but admit you haven't made use of it and
remain ignorant.

as it just doesn't make sense] is that White Balance information is stored as
encrypted metadata in the RAW file. It is NOT applied until conversion time.

If you don't understand even the *basics* of underlying
hardware, how can you claim to know what a digital camera is or
is not doing? As I noted, saying there is no analog level is
indicative. (And your discussion is indeed following what that
indicates...)

Now you say there is no reason "to believe otherwise as it just
doesn't make sense" to something that has been pointed out to
you as one of the more significant technical features of Nikon's
recent DSLR models.

Multi-channel i/o with gain adjusted amplifiers between the
sensor and a 4 channel analog to digital converter is rather
obvious as a way to increase the speed of operation, and adding
white balance at that point is a little bit of frosting on the
cake to make it even faster.

It clearly makes *great* sense! But you have to understand what
the analog signal between the sensor and the AD converter is
too... and denying that it exists is not a good starting point!

"As you can see from the diagram above the D2X reads data
from the sensor using four channels. To keep color
information as pure as possible this is done by grouping
pixels together by color; R, Gr, B, Gb. This four channel
readout is then fed to the analogue processing system / A/D
converter as separate channels and is processed as such. This
allows the camera to achieve its five frames per second
continuous shooting rate.

I didn't argue anything about 4-channels. I argued that White Balance
information is NOT applied in camera to a RAW file. A/D conversion is just
counting photons [and noise].

Wrong on all points other than you didn't say anything about
4-channels. Not knowing about the hardware is exactly why you
are making incorrect sweeping statements about what the hardware
can or cannot do.

"The diagram above clearly demonstrates the D2X's four
channel readout and independent color processing and A/D
conversion which is carried out. Although Nikon aren't
revealing a huge amount of detail about what goes on at the
'pre-conditioning' stage (just before A/D) the idea is that
data from the individual channels can be amplified
specifically to apply some amount of WB correction and to
maximize color gradations before conversion to a digital
value in the A/D converter.

The diagram is poor, and does not indicate what the text describes.

In fact, that text specifically says that they do not know what happens in
"pre-conditioning" and simply surmise that some amount of White Balance

It says Nikon has said that is what it does, but has not
indicated exactly *how* it is done. The general principles are
well enough understood, but Nikon is not revealing specifics.

I did provide you with information on how to research this, so I
don't see why you continue to make such statements. Pulled from
the same google search that I suggested, see

http://www.bythom.com/d200review.htm

Where Thom Hogan describes the Nikon D200 hardwa

"... data is manipulated in the analog space (electrons)
prior to amplification and digital conversion. This is most
prevalent where individual channels are amplified prior to
the ADC to adjust white balance, ..."

information is applied. Clearly ... that remains a mystery. The commonly
heald believe, from what I have read is that it is not.

From what you have read? Do you have a reference that says
Nikon is trying to deceive people with the data they have
released?

It can easily be demonstrated that results from the D2x are
consistent with White Balance being done in hardware prior to
A/D conversion.

A 16 bit TIFF formatted file is easily able to contain all of
the data available in a 12 bit NEF file.

I indicated that they would be similarily sized files for the same bit-depth.
They would.

They won't. A D2x produces a 4312x2868 pixel image, which
contains 12.37 MP. The file size if the non-compressed NEF mode
is selected will be roughly 20Mb per image. Converted to 16 bit
TIFF format that will produce a 37Mb file. If compressed it
will be slightly smaller 30-35Mb.)

I mis wrote there, as I meant to say analgous bit-depths. Clearly, a 12-bit
RAW is NOT the same thing as a 12-bit TIFF.

But it *does* contain all the same data. Others have provided
details of the file size differences, so I'm not going to repeat
that here.

--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska)

Floyd L. Davidson wrote:

http://www.bythom.com/d200review.htm

Where Thom Hogan describes the Nikon D200 hardwa

"... data is manipulated in the analog space (electrons)
prior to amplification and digital conversion. This is most
prevalent where individual channels are amplified prior to
the ADC to adjust white balance, ..."


Floyd,

If Thom Hogan says that then he's wrong. I checked this once upon a
time (on a rainy, windy day when I had nothing else to do on my D200
by photographing a gray card with my camera on a tripod and changing
only WB (from the max temperature to the min, whatever they are), then
looking at raw data in IRIS (I also checked in normal converters, but
used IRIS to make doubly sure). The raw data doesn't change (at least,
it doesn't change more than it changes between successive exposures
with the same WB).

It is true that Nikon go on in the D200 brochure about how individual
channel "preconditioning" is done in the D200 (I don't think I've seen
any explicit claim that this "preconditioning", whatever it is, changes
with WB setting). So I suppose what they do is something like amplify
each channel indepently before the ADC, but by fixed amounts,
independent of the WB setting (maybe to bring each channel closer to
equality with the others assuming daylight, or something like that). Or
they're lying, but I don't think they would (what for?).

But this is about the D200, probably it's different in the D2X (and I
do remember they were claiming it does what you say; between that and
you saying you checked it, I have no reason to doubt it).

"acl" wrote:
Floyd L. Davidson wrote:

http://www.bythom.com/d200review.htm

Where Thom Hogan describes the Nikon D200 hardwa

"... data is manipulated in the analog space (electrons)
prior to amplification and digital conversion. This is most
prevalent where individual channels are amplified prior to
the ADC to adjust white balance, ..."


Floyd,

If Thom Hogan says that then he's wrong. I checked this once upon a
time (on a rainy, windy day when I had nothing else to do on my D200
by photographing a gray card with my camera on a tripod and changing
only WB (from the max temperature to the min, whatever they are), then
looking at raw data in IRIS (I also checked in normal converters, but
used IRIS to make doubly sure). The raw data doesn't change (at least,
it doesn't change more than it changes between successive exposures
with the same WB).

It is true that Nikon go on in the D200 brochure about how individual
channel "preconditioning" is done in the D200 (I don't think I've seen
any explicit claim that this "preconditioning", whatever it is, changes
with WB setting). So I suppose what they do is something like amplify
each channel indepently before the ADC, but by fixed amounts,
independent of the WB setting (maybe to bring each channel closer to
equality with the others assuming daylight, or something like that). Or
they're lying, but I don't think they would (what for?).

But this is about the D200, probably it's different in the D2X (and I
do remember they were claiming it does what you say; between that and
you saying you checked it, I have no reason to doubt it).

Interesting! Per the discussion you and I had a couple weeks
ago about histogram generation, the results that I got with the
D2x are consistant with WB being done prior to quantization.

Now the question would be if Hogan misunderstood something Nikon
said about the D200, or if Nikon provided information that was
misleading. (He probably got confused with the D2x...)

--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska)

Floyd L. Davidson wrote:

Interesting! Per the discussion you and I had a couple weeks
ago about histogram generation, the results that I got with the
D2x are consistant with WB being done prior to quantization.

Yes, I think you had said that changing WB was the only thing that
changed the histogram (ie saturation, sharpness etc did not)? If so, I
suppose it's possible that either the histogram is not made from the
small jpeg embedded in the raw file, or it is made from the jpeg but
that jpeg itself is unaffected by sat, colour space etc settings.
Either way, it's different than for other cameras I have heard about.

Now the question would be if Hogan misunderstood something Nikon
said about the D200, or if Nikon provided information that was
misleading. (He probably got confused with the D2x...)


Possibly Nikon wanted to use manipulations in analog space for WB but
decided against it to avoid confusing people. I wouldn't be surprised.

(Floyd L. Davidson) writes:

Interesting! Per the discussion you and I had a couple weeks ago
about histogram generation, the results that I got with the D2x are
consistant with WB being done prior to quantization.

That is because the histogram is computed from the preview jpg, not
from the real, raw, data. Trying to do white balance by changing the
preamp gain would be insane, you would have to change gain settings
for every pixel you read down! And keep it all stable.

--
Paul Repacholi 1 Crescent Rd.,
+61 (08) 9257-1001 Kalamunda.
West Australia 6076
comp.os.vms,- The Older, Grumpier Slashdot
Raw, Cooked or Well-done, it's all half baked.
EPIC, The Architecture of the future, always has been, always will be.
------------ And now a word from our sponsor ----------------------
For a quality mail server, try SurgeMail, easy to install,
fast, efficient and reliable. Run a million users on a standard
PC running NT or Unix without running out of power, use the best!
---- See http://netwinsite.com/sponsor/sponsor_surgemail.htm ----

writes:
That is because the histogram is computed from the preview jpg, not
from the real, raw, data. Trying to do white balance by changing the
preamp gain would be insane, you would have to change gain settings
for every pixel you read down! And keep it all stable.

Not separate amps for R,G,B?

wrote:
(Floyd L. Davidson) writes:

Interesting! Per the discussion you and I had a couple weeks ago
about histogram generation, the results that I got with the D2x are
consistant with WB being done prior to quantization.

That is because the histogram is computed from the preview jpg, not
from the real, raw, data.

There is no difference between computing it from the "preview
jpg" and the "real" raw data, as that would be the same either
way. The point was that on some cameras it appears to be
computed from the output JPEG file, and various in camera
adjustments to that file (sharpening, saturation, etc.) will
affect the histogram even if the camera is in "RAW mode".

With the D2x none of the adjustments that affect the JPEG will
affect the histogram. But WB does.

Nikon has stated that they do use analog gain control, before
the A/D conversion. But they have not been overly generous in
describing exactly what they do, so we don't know. It does
sound as if the entire WB might not be done at the analog level,
and some of it might be after digitization. Regardless,
whatever it is they do it does have to be part of the data saved
with the RAW file, for post processing.

Trying to do white balance by changing the
preamp gain would be insane, you would have to change gain settings
for every pixel you read down! And keep it all stable.

I see little difference between changing gain once per image or
3.1 million times per image. That's pretty slow as far as
signal processing goes.

--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska)

Paul Rubin wrote:
writes:
That is because the histogram is computed from the preview jpg, not
from the real, raw, data. Trying to do white balance by changing the
preamp gain would be insane, you would have to change gain settings
for every pixel you read down! And keep it all stable.

Not separate amps for R,G,B?

The Nikon D2x (and the D200, though it appears it does *not* do
analog gain control) has 4 channels between the sensor and the
4-channel digitizer, R, G, B, and another G.

Most sensors have used a 2 channel layout.

--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska)

Paul Rubin writes:

writes:
That is because the histogram is computed from the preview jpg, not
from the real, raw, data. Trying to do white balance by changing the
preamp gain would be insane, you would have to change gain settings
for every pixel you read down! And keep it all stable.

Not separate amps for R,G,B?

No, the data comes out like RGRGRG.... GBGBGBGB

All through the same (set) of preamps. With a CMOS chip you *could*
wire up seperate gain controls, but I suspect you dont want that
much extra metal over your chip.

--
Paul Repacholi 1 Crescent Rd.,
+61 (08) 9257-1001 Kalamunda.
West Australia 6076
comp.os.vms,- The Older, Grumpier Slashdot
Raw, Cooked or Well-done, it's all half baked.
EPIC, The Architecture of the future, always has been, always will be.