By the numbers ? and the D1


Andrew Rodney <andrew@...>
 

on 1/2/01 10:35 AM, Rob Outlaw at routlaw@imt.net wrote:

At any rate their claim was that "Nikon did not embed this
profile but only designed the NTSC working space as a color characteristic".
Which is fine IF that's really the case. I don't need an embedded profile
(it's nice). Having an untagged file that I KNOW is in a certain space and
having a profile to now assign to that file is all I need. As soon as I
Assign this profile, the preview in Photoshop 6 alters based on this new
information. The image looks really good! And when I convert to Adobe RGB,
it's a correct conversion because the source used to get to Adobe RGB is
correct. If I assume a different space, the conversions as well as the
previews undergo some degree of being hosed. So again, the question is, is
the file off the D1 really in NTSC space?

since Nikon does not let the general public in on what they really are
thinking, my assumptions would be that they have somehow within their RGB
capture used the x & y color coordinates of the NTSC working space as the color
gamut for the raw capture.
The real scary part here is what you said about Nikon not letting the
general public in on the color the unit produces. What are they thinking? If
you are going to force the non raw files into a colorspace, why NTSC? At
least with the S1 Pro, assuming ColorMatch RGB produces a very acceptable
preview because the file is massaged towards this space. If you don't use
Photoshop 5 or 6, the image still looks pretty good in non color managed
products. At least they didn't go sRGB.

So with that in mind and for those having trouble with flesh tones with the
D1, I still argue for
a total elimination of NTSC from a given workflow, since it does have an
extended red point
on the color map.
Actually IF the files really were in NTSC RGB and assigning this profile
produced good previews and conversions, I'd be OK with it. I question
whether the files are in NTSC RGB as described in the Working Space of that
name in Photoshop 5 or not.

Taken a step further anytime that I have completed a P2P
from say
NTSC to Adobe RGB or Colormatch RGB (while working in that space) reds do
become
more saturated and it is very clear to see this on my monitor, again
exacerbating any problems
that exist with overly red/magenta flesh tones.
Photoshop 5? You are changing your Working Space to now match up with the
file right? The previews should NOT change a lick. In Photoshop 6 they don't
because of the Document Specific Color and new preview pref's. In Photoshop
5 the preview would change after PtoP but as soon as you loaded that profile
you used in the PtoP, the preview would change back (and match) what you saw
prior to PtoP. Photoshop 6 makes all this so much easier!

Fundamentally I can not argue with this, makes perfect sense, but
unfortunately
probably the vast majority of D1 shooters out there do not have an adequate
custom profile, but have relied on Nikon to provide them with an out of the
box
camera with perfect color. I think it is safe to say Nikon let them down.
Nikon (and all manufacturers) have only a few options:

1. Pick a "common" RGB space and despite the gamut abilities of my sensor,
funnel the color into this space. In the case of the Nikon, we are told it's
NTSC RGB (I don't necessarily buy that but perhaps). Others could pick sRGB.
The upside is we can define the color without doing a thing (assuming the
color really does get funneled to that space). Downside is you get funneled
data which for some uses is just fine and for other cases not good.

2. Pick a non common, non profiled RGB. Now things get dicy. We don't have a
clue what the color really is. Yet in Photoshop, we have to tell it
something about the color of the file so we can preview it and make
conversions. We don't have a profile so we have to guess. Guess the RGB is
the name of this game. Pick one that's close to the reality of the file and
you'll do pretty well. Pick on that's not close and you'll get all kinds of
nasty previews and output. That magenta skin stuff for the D1 is an example.

3. Pick a non common but profiled space. So you buy a PhaseOne back and you
get a bunch of camera profiles. Are they any good? Not bad. At least we are
somewhat dialed into the color of the chip or the color the software gets.
Better yet, make a custom profile.

Andrew Rodney


Rob Outlaw <routlaw@...>
 

Andrew wrote:

If indeed this is true on Nikons part, they dont need to embed a profile
(be nice). We only need to assign the NTSC profile to the data provided
for
conversion into another space. FWIW, the gamut map I have of the D1 from a
custom profile doesnt look to me like NTSC!
Perhaps I should elaborate on what Nikon claims if that is possible. This
info came
from the D1 discussion list moderated by Juergen Specht, and was a reply not
only
from Bill Pekala but also Stephen Pont of Nikon back when they were
responding
on that digest. At any rate their claim was that "Nikon did not embed this
profile but
only designed the NTSC working space as a color characteristic". Your guess
is as good
as mine as to what this really means technically, but as long as we have to
assume things
since Nikon does not let the general public in on what they really are
thinking, my
assumptions would be that they have somehow within their RGB capture used
the
x & y color coordinates of the NTSC working space as the color gamut for the
raw
capture.

So with that in mind and for those having trouble with flesh tones with the
D1, I still argue for
a total elimination of NTSC from a given workflow, since it does have an
extended red point
on the color map. Taken a step further anytime that I have completed a P2P
from say
NTSC to Adobe RGB or Colormatch RGB (while working in that space) reds do
become
more saturated and it is very clear to see this on my monitor, again
exacerbating any problems
that exist with overly red/magenta flesh tones.

That's not an indication alone we have an issue with reds only that the
camera has a wide gamut falling into reds. Again, with a proper profile
assigned, there is no red/magenta issue. In the gamut map I have of the
D1,
the blues fall off the CIE chart!
Fundamentally I can not argue with this, makes perfect sense, but
unfortunately
probably the vast majority of D1 shooters out there do not have an adequate
custom profile, but have relied on Nikon to provide them with an out of the
box
camera with perfect color. I think it is safe to say Nikon let them down.

I hope I do not sound like a broken record here, one can get really good
color
without the profile and with a few minor adjustments in PS with the raw
files at
least, but only if eliminating the NTSC issue. That has been my experience
with
the camera.

> I suspect when you use the Nikon Capture (shame they have to charge for
it),

Don't even get me started on this one! While I use it all the time it has to
be the
worst $450.00 I have ever spent in photography.

you may not get NTSC but indeed the raw RGB. I'm guessing that when using
other than RAW, Nikon *may* be forcing the raw RGB into NTSC although I
suspect this is may not be that accurate but convenient for Nikon to
simply
suggest to users.
If this were true with non raw data files when opened in PS 5.5 or later
would I not
be asked about converting from NTSC to Adobe or whatever? I can't think of
once
that this has happened for me, and I do have PS setup so that would happen
if a
profile were embedded or attached.

Of all the companies making digital cameras and scanners,
Nikon's track record for dealing with color issues is pretty bad.
I do not doubt this for a second.

Rob


Andrew Rodney <andrew@...>
 

on 1/2/01 6:59 AM, Rob Outlaw at routlaw@IMT.NET wrote:

Actually as I understand it Nikon has only "chosen the NTSC working space as
a color characteristic" but has not embedded, tagged, or otherwise forced
the issue
of using this as a working space or profile.
If indeed this is true on Nikon¹s part, they don¹t need to embed a profile
(be nice). We only need to assign the NTSC profile to the data provided for
conversion into another space. FWIW, the gamut map I have of the D1 from a
custom profile doesn¹t look to me like NTSC!

Nikon has forced the issue. The camera is producing some kind of RGB. I¹d be
inclined to always use the RAW mode of capture to get to that color. As Bob
pointed out, the Kodak DCS line of cameras is much more savvy when it comes
to dealing with raw captures. The Linear 12 bit capture is ideal for
profiling and truly does provide the user with a raw color file. It looks
just awful when you bring it into Photoshop and you are viewing it with some
preset Working Space. But it¹s untagged and as soon as you Assign the proper
profile (easy to do in Photoshop 6) the preview instantly changes and looks
fantastic. A further illustration of how viewing a file without a proper tag
can be a recipe for disaster!

You are so right here Dan. If one looks at the color points on a color gamut
map it is clear to see that the NTSC working space has a red point that is
virtually passed human vison and certainly passed Adobe RGB let alone
Colormatch.
That's not an indication alone we have an issue with reds only that the
camera has a wide gamut falling into reds. Again, with a proper profile
assigned, there is no red/magenta issue. In the gamut map I have of the D1,
the blues fall off the CIE chart!

In not one of those assignments did the film
outperform
the color of my D1, capturing images in raw mode and converting with Nikon
Capture,
while totally ignoring the NTSC working space. All files were opened in PS
5.5 into the
Adobe RGB space without any other conversions taking place and saved as
such.
I suspect when you use the Nikon Capture (shame they have to charge for it),
you may not get NTSC but indeed the raw RGB. I'm guessing that when using
other than RAW, Nikon *may* be forcing the raw RGB into NTSC although I
suspect this is may not be that accurate but convenient for Nikon to simply
suggest to users. Of all the companies making digital cameras and scanners,
Nikon's track record for dealing with color issues is pretty bad.

Andrew Rodney
Andrew Rodney


Bob Smith <rmsmith@...>
 

Rob Outlaw wrote:

D1, capturing images in raw mode and converting with Nikon Capture, while
totally ignoring the NTSC working space.
which is exactly how it ought to be done. NTSC is just an issue for camera
processed files. My experience with D1 jpegs is nil... just going by
experiences of others that I trust and my very few efforts at trying to help
someone with some D1 jpegs. Except possibly on some studio type rigs, I
don't think any digital camera images from any company are tagged coming out
of the camera. It makes no sense to waste precious writing speed and disk
space on data that would be exactly the same for every image... especially
on a camera who's main claim to fame is photojournalistic use.

Bob Smith


Rob Outlaw <routlaw@...>
 

Bob writes,
Its not a matter of being wider. Its a matter of matching what Nikon was
targeting their output to. Why, I don't know... but Nikon has said that
Nikon D1 in camera processing is done with output to NTSC rgb in mind. It
has to be going to some definition of RGB and that's what Nikon chose.>>
Actually as I understand it Nikon has only "chosen the NTSC working space as
a color characteristic" but has not embedded, tagged, or otherwise forced
the issue
of using this as a working space or profile.

Dan writes:

I have played with a D1 but don't claim any expertise in it. What I can
say
for sure is, if the problem is that fleshtones are too magenta, then
assuming that the files are NTSC is going to make that problem worse, not
better.
You are so right here Dan. If one looks at the color points on a color gamut
map it is clear to see that the NTSC working space has a red point that is
virtually passed human vison and certainly passed Adobe RGB let alone
Colormatch.
By utilizing the NTSC as part of your workflow
whether converting from or into regardless of the space that you do this
from,
will only exacerbate the red flesh tone problem. I have spent literally
dozens
of hours experimenting with this situation. My conclusions are that totally
ignoring NTSC in any shape form or fashion is a better solution at least for
flesh tones anyway.

Over the last year I have had the opportunity to shoot some four or five
assignments
where the decision was made to shoot D1 images alongside film (in this case
Astia and
Provia F my films of choice). In not one of those assignments did the film
outperform
the color of my D1, capturing images in raw mode and converting with Nikon
Capture,
while totally ignoring the NTSC working space. All files were opened in PS
5.5 into the
Adobe RGB space without any other conversions taking place and saved as
such.
I am not saying that some minor tweaking of the color in some cases was not
needed and
most of that was done in the Hue and Sat window where I could selectively
edit colors
(mainly red channel for flesh tones and the yellow channel for overly green
yellows)
in 16 bit mode.

Having said all that I will admit to the fact that this situation for bad
flesh tones
with the D1 seems to be worse for jpeg files as opposed to raw file capture.
I have
no idea why this is the case either. However I am not sure it is worse for
outdoor
shots as compared to on camera flash, but the placement of a UV filter over
the flash unit does help while not completely eleminating the overly
red/magenta
flesh tones.

Rob Outlaw