Date: Fri, 26 Apr 2002 16:55:18 -0000
From: "dave_cardinal"
Subject: That darn EXIF profile Tag:-)
Since there has been so much written about what the cameras are doing with the EXIF Colorspace tag I thought folks might be interested in the actual definition of the tag. From the EXIF2.1 spec:-> Quote from spec... ->
ColorSpace:
The color space information tag (ColorSpace) is always recorded as the color space specifier. Normally sRGB (=1) is used to define the color space based on the PC monitor conditions and environment. If a color space other than sRGB is used, Uncalibrated (=FFFF.H) is set. Image data recorded as Uncalibrated can be treated as sRGB when it is converted to FlashPix.
...
Tag = 40961 (A001.H) ...
1 = sRGB
FFFF.H = Uncalibrated
Other = reserved
[- End Quote<-Aside from being somewhat confusing as written, it also has the obvious defect of not allowing any direct specification of a calibrated image other than sRGB.
For the camera companies, it is fairly easy to set this tag one way or the other (and clearly cameras not using sRGB should set it to uncalibrated) but it is a whole different ballgame for them to embed a profile in the image. That requires more than just another tag, it requires some additional logic to write a new APP segment, etc.
In particular, to assign any custom profiles in the camera would be a major change as they tend to be quite large, affecting image size and requiring multiple segments in the file. The best current solution we've found is to allow our users to specify which profiles they'd like our software to assign to which camera/colorspace combinations as the images are loaded into the computer.
--David Cardinal
Pro Shooters LLC
http://www.proshooters.com
Date: Fri, 26 Apr 2002 15:03:28 -0600
From: Andrew Rodney
Subject: Re: That darn EXIF profile Tag:-)
on 4/26/02 10:55 AM, dave_cardinal at yahoo@cardinalfamily.com wrote:> In particular, to assign any custom profiles in the camera would be a
> major change as they tend to be quite large, affecting image size and
> requiring multiple segments in the file.
>
Kind of funny when you see that most consumer cameras are trying to produce bigger and bigger files (remember with a mega-pixel capture was big)? The profiles are rather small in comparison. For pro capture devices, 5 mega-pixel files seem to be the starting point. Custom profile I1ve build to for cameras are under 100K!> The best current solution
> we've found is to allow our users to specify which profiles they'd
> like our software to assign to which camera/colorspace combinations
> as the images are loaded into the computer.That is the best solution, no question. If only more manufacturers would do this!
Andrew Rodney
Date: Sun, 28 Apr 2002 14:20:39 -0600
From: Chris Murphy
Subject: Re: That darn EXIF profile Tag:-)
dave_cardinal (yahoo@cardinalfamily.com) writes:>Aside from being somewhat confusing as written, it also has the
>obvious defect of not allowing any direct specification of a
>calibrated image other than sRGB.They need a longer list, 2 could be ColorMatch RGB, 3 could be Adobe RGB (1998), and so on. That way those profiles don't have to be explicitly embedded in every single image.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Tue, 30 Apr 2002 16:42:46 -0000
From: "mrmattbarton"
Subject: PS 7 profiles & the D1x trouble
Hi-
We finally got PS 7 in the studio today. After a few tests, we started checking the EXIF info from our Nikon D1x camera files. We always set the D1x color space to Adobe RGB (1998). But according to EXIF info in PS 7, the D1x is shooting in sRGB regardless of the setting. This causes the color profile warning in PS 7 to come up before EVERY file we open. Ideally, we'd like the Adobe RGB (1998) profile to carry over from our camera to our computers. Has anyone else experienced this? Is this a Nikon problem?I'm on a G4/400 in 10.1.3. Thanks for any answers.
Matt
Date: Tue, 30 Apr 2002 15:14:38 -0500
From: Bob Smith
Subject: Re: PS 7 profiles & the D1x trouble
mrmattbarton wrote:> Ideally, we'd like the Adobe RGB (1998) profile to carry over from our camera
> to our computers.When you set the camera for Adobe RGB I'm betting that it doesn't embed the profile in camera produced JPEGS, it just uses that space for output. It makes little sense to waste precious camera processing power and disk space to write the same chunk of data (even a very small one) to every image...especially with a camera that prides itself on shooting speed. With no real profile present, it sounds like PS7 is now honoring the EXIF data where PS6 didn't. If a real profile were there, its my understanding that it takes precedence over the EXIF tag.
Bob Smith
Date: Tue, 30 Apr 2002 17:40:49 -0600
From: Andrew Rodney
Subject: Re: PS 7 profiles & the D1x trouble
on 4/30/02 10:42 AM, Matt Barton wrote:> Has anyone else experienced this? Is this a
> Nikon problem?
Sure is. They (Nikon) should be embedded Adobe RGB. The EXIF data that Nikon places in the file is wrong.Andrew Rodney
Date: Thu, 02 May 2002 13:49:23 -0000
From: "mrmattbarton"
Subject: Re: PS 7 profiles & the D1x trouble
Generally I only shoot Jpegs so I never bothered to install Nikon Capture. But if it's as good as you say, I'll give it a shot.Someone from Galbraiths forums suggested I check my colorsync settings. Which I did and nothing changed. It appears the D1x is using the AdobeRGB(1998) color space but can't communicate that to PS 7. So my question is...
When I tell PS 7 to convert a new image to my workspace profile (AdobeRGB) does it really change anything? Assuming the image is already (secretly) AdobeRGB.
Date: Thu, 2 May 2002 12:44:08 -0600
From: Chris Murphy
Subject: Re: Re: PS 7 profiles & the D1x trouble
Matt Barton writes:>When I tell PS 7 to convert a new image to my workspace profile
>(AdobeRGB) does it really change anything? Assuming the
>image is already (secretly) AdobeRGB.Anytime you use Convert to profile and the source and destination profiles are different, YES a conversion is occuring. What you want to do instead of Convert to Profile is use Assign Profile and assign Adobe RGB
(1998) which will not convert the image.Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Fri, 03 May 2002 01:46:51 -0000
From: "denkjersey"
Subject: profiles and Photoshop 7
I'm brand new here, so please excuse me if this has been discussed before. With Photoshop 6.0 my Fuji S-1 camera files opened as "untagged" which gave me the option to assign my custom camera profile (created using Andrew Rodney's technique..thanks Andrew) and then convert to my standard working space. Photoshop 7 however, now "sees" my camera files as tagged with the srgb profile which is gives me no where near as good color as my custom profile. It's a huge pain having to open files, click on "don't color manage" assign my profile and then convert to the working space...plus..I don't want to jpeg these files again...so I'm saving them as tiffs with the new color space embedded..which is taking up huge amounts of space. Does anyone know a way to tell Photoshop not to assign srgb to my camera files? Thanks in advance.
Dennis Kelly
Date: Fri, 3 May 2002 09:22:36 -0600
From: Chris Murphy
Subject: Re: profiles and Photoshop 7
denkjersey (dennis.p.kelly@verizon.net) writes:
>Does anyone know a way to tell Photoshop not to assign srgb to my
>camera files?No. Call Fuji, tell them that they are improperly assigning sRGB as the profile in EXIF data in your files. If you don't complain, it won't get fixed, and the problem is primarily with the digital camera vendors, not Adobe. Although Adobe it being derelict for not coming up with a way to deal with this reality (actually they had a way, but broke it in Photoshop 7).
What I recommend doing is make a Droplet action in Photoshop that you can drag an entire folder of images on, and it will remove sRGB, reassign the profile you want, then convert to your working space and resave the image. Once it's created, it will be a much quicker way of dealing with the issue.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Fri, 03 May 2002 12:19:02 -0400
From: Ric Cohn
Subject: Re: profiles and Photoshop 7
Many (but not all) of us feel your pain. You need to open the file without conversion and then do an assign profile to the correct camera profile and resave.Unless people are posting crank questions to this site, I think the question of whether Photoshop's new behavior is a bad idea has been answered by the users!
Ric Cohn
ric@riccohn.com
http://www.riccohn.com
212.924.4450>> Photoshop 7 however, now "sees" my camera files as tagged with the
> srgb profile which is gives me no where near as good color as my
> custom profile. It's a huge pain having to open files, click
> on "don't color manage" assign my profile and then convert to the
> working space...plus..I don't want to jpeg these files again...so I'm
> saving them as tiffs with the new color space embedded..which is
> taking up huge amounts of space.
> Does anyone know a way to tell Photoshop not to assign srgb to my
> camera files?
Date: Fri, 3 May 2002 11:14:01 -0600
From: Chris Murphy
Subject: Re: profiles and Photoshop 7 Thanks!
Dennis Kelly (dennis.p.kelly@verizon.net) writes:>Chris..thanks for the quick reply. I'll have to read up on droplets.
>One quick question...if I open a jpeg, change only the color profile and
>save it, (as a jpeg) am I recompressing and losing color information, or is
>the image data left unchanged and just the exif data altered?I think as long as you don't do a Save As, and the change only occurs to non-image data, there is not lose of information (artifacting) as there is no need for the image to be rerun through the JPEG algorithm. At least my file sizes aren't changes substantially (a few bytes) when I add and remove guides and do just a save. In the past, if I do a save as though, file sizes start to drop rapidly indicating to me compression is throwing out more data. However, I just tried this in Photoshop 7 and the file size is the same - so it's almost like it knows that there is no change in the image portion, or that given the quality setting, the data is compressed as much as it should be.
I'm really not sure :)
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Sat, 04 May 2002 12:56:01 -0000
From: "rgb2cmykuk"
Subject: Re: profiles and Photoshop 7
--- In colortheory@y..., Chris Murphy wrote:
> What I recommend doing is make a Droplet action in Photoshop that you can
> drag an entire folder of images on, and it will remove sRGB, reassign the
> profile you want, then convert to your working space and resave the
> image. Once it's created, it will be a much quicker way of dealing with
> the issue.Couldn't work out what platform the original poster was using, but if it's Mac then the easiest thing to do is drag and drop your images on the free Applescripts that are supplied with the OS.
Just drag all of the images on the "Embed chosen profile" script and then choose a profile. Alternatively, if ColorSync is correctly set up, you can "Embed specific profile" or "Embed display profile".
This will be a whole lot quicker than getting PS to open and resave all of your images. And if you wanted to you could easily adapt the scripts to re-profile all of the images that you drop into a hot folder.
--
Martin Orpen
Idea Digital Imaging Ltd -- The Image Specialists
http://www.idea-digital.com
Date: Sat, 4 May 2002 08:42:17 -0700
From: Jan Steinman
Subject: Re: profiles and Photoshop 7
From: Chris Murphy>What I recommend doing is make a Droplet action in Photoshop that you can
>drag an entire folder of images on, and it will remove sRGB, reassign the
>profile you want...If I may be so brash as to potentially start a platform war...
MacOS has had droplets for changing profiles for nearly a decade. That may be why many MacOS users see this (and many other so-called "problems" that Windows users have when doing graphical stuff) as a non-issue.
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Sun, 5 May 2002 12:19:01 -0600
From: Chris Murphy
Subject: Re: Re: profiles and Photoshop 7
Jan Steinman (Jan@Bytesmiths.com) writes:>MacOS has had droplets for changing profiles for nearly a decade.
This is true, and they are quite useful.
>That may
>be why many MacOS users see this (and many other so-called "problems" that
>Windows users have when doing graphical stuff) as a non-issue.I wouldn't go that far on the color management front - the ColorSync AppleScripts alone aren't something most users realize is installed by default. I think if anything Mac users have more of these problems in terms of quantity just because there are so many more of them in this industry.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Mon, 20 May 2002 16:59:50 -0400
From: Scott Olswold
Subject: EXIF, sRGB, and Embedded Profiles
RE: The Microsoft monopoly on the "default" RGB working space.
I'm not going to stamp my foot in the dirt and say that Microsoft is explicity causing the sRGB debacle regarding the EXIF tag for the color space. That this might be happening implicitly is another matter entirely. A digital camera manufacturer has to answer a bunch of questions regarding its product. The first is "Who is using my digital camera?" For most cameras (<$500 US), this will end up being a bunch of moms and dads, and professional photo types who want a kick-around cam for the occasional outing. And with the exception of the pro, they'll be on a Windows PC with an uncalibrated monitor. And they'll be printing on photo-quality inkjets. So the sRGB default makes some sense. I really think this should be separated between 'consumer' models (for those of us that color correct and tweak in 8-bit RGB<g>) and 'professional' models--with consumer models defaulting to this "sorta-kinda-maybe" sRGB behavior (but having this explained distinctly in the User Guide--oh, wait, nobody reads those), but being able to change the behavior, and the professional models offering the mechanism to define their behavior from the beginning. Since EXIF is so extensible, it shouldn't be a big deal, and some of our standard RGB spaces (plus "raw") shouldn't suck up too much space on the camera's ROM.
To that end, if the default behavior *is* indeed reported as sRGB, then it darn well should *be* sRGB. That's a problem that needs to be fixed on the manufacturer's side.
RE: Andrew's policy of tinkering with files with the embedded sRGB tag.
I agree with Chris that this stands to put the concept of embedded profiles on its ear. I have come to trust Andrew on many different topics (particularly when there's concensus between several of you whom I refer to as my mentors), but this one I disagree with. Maybe this is in response to Photoshop 5's insistance on defaulting to the sRGB working space, I don't know. But if the user bothered to calibrate the monitor, then even this was accurate, yes? I mean, as far as assigning meaning to the colors. That this may behave poorly in press, is another matter entirely (and thus up for realignment)...but second guessing is bad in the long run. My first reaction is to trust the embedded profile until soft-proofing shows me an awful mess (or the original RGB is an awful mess).
But like editing in 16 or 8-bit...diff'rent strokes.
Scott Olswold
Date: Mon, 20 May 2002 15:38:52 -0600
From: Chris Murphy
Subject: Re: EXIF, sRGB, and Embedded Profiles
Scott Olswold (scotto@mail.digitaldanka.com) writes:>For most cameras
>(<$500 US), this will end up being a bunch of moms and dads, and
>professional photo types who want a kick-around cam for the occasional
>outing. And with the exception of the pro, they'll be on a Windows PC with
>an uncalibrated monitor. And they'll be printing on photo-quality inkjets.
>So the sRGB default makes some sense.sRGB is not the worst thing in the world, and for process printing it really isn't going to matter than much anyway. The problem is that they say it behaves per sRGB, but it behaves like something else. That's a problem for consumers and professionals alike.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
on 5/20/02 2:59 PM, Scott Olswold at scotto@mail.digitaldanka.com wrote:> I agree with Chris that this stands to put the concept of embedded profiles
> on its ear. I have come to trust Andrew on many different topics
> (particularly when there's concensus between several of you whom I refer to
> as my mentors), but this one I disagree with. Maybe this is in response to
> Photoshop 5's insistance on defaulting to the sRGB working space, I don't
> know. But if the user bothered to calibrate the monitor, then even this was
> accurate, yes? I mean, as far as assigning meaning to the colors. That this
> may behave poorly in press, is another matter entirely (and thus up for
> realignment)...but second guessing is bad in the long run. My first reaction
> is to trust the embedded profile until soft-proofing shows me an awful mess
> (or the original RGB is an awful mess).We have three conditions with regard to files and embedded profiles or tags:
1.The correct profile. The file should look great on my calibrated display. That is, if I1m in the position where I1m going to edit the file based on what I want, then the file should look good. If I1m in the position where I1m only supposed to open and print the file (my judgment on how the file looks isn1t material) then I1m not going to mess with the numbers anyway.
2. The profile is wrong. Same as above but I have to decide if I like what I see and I1m allowed to alter the file. I1ve got a calibrate display. The file looks bad. The profile may actually be 3right2 but the file looks bad. I can improve the appearance by changing the embedded profile. Why not? Am basically saying to the CMS, 3I like what I see, now match it2 (naturally we have issues like soft proofing to the output space, rendering intents and so forth. But you get the idea that all I want is the file to look the way I want it to look). The person who provided me the file may have had a display that was wacked out. Or the user might have been a bone head and embedded the wrong profile. Or some manufacturer added a tag that isn1t at all correct. I have to fix the issue. I don1t like the idea that the profile is wrong but the bigger issue is to make the darn file look good and print that way.
3. No profile. Now I have to make all kinds of guesses. But the positions I take are the same as above. I am supposed to make the file look good and move on. But I have a lot of guessing to do.
You1ve got to calibrate the display in any of the cases above. Photoshop is going to produce a preview I can look at and decide if I like using that calibrated display and the embedded profile (be it correct or not).
I1d prefer that EVERY file have the right embedded profile. But I may decide based on my role here (which is to alter the file), I don1t like what I see. I can alter the pixels themselves or I can alter the meaning of the numbers and perhaps get a bit closer to my goal. Or I can do both. But if I get a file that1s really butt ugly and has an embedded profile, I1m going to assign a few different profiles and simply look at the display and see what my options are. Doesn1t hurt the file. Doesn1t take any time (even on a huge file). Might help.
Andrew Rodney
Date: Mon, 20 May 2002 22:39:06 -0700
From: Rick Gordon
Subject: Re: EXIF, sRGB, and Embedded Profiles
The one element of this discussion that I don't believe I've seen yet, and perhaps the easiest way for Adobe to satisfy people on both sides of this argument is to add to the interface of the opening Profile Mismatch dialog:1) The option to assign a different profile should be available in that dialog as the file is opened, as it is for untagged files.
2) Secondly, Adobe should add an option for saving a default behavior for reassigning EXIF profile data, so that it can be handled automatically from that point (which may be primarily useful for individuals who are importing from a single device, but that would be the case for many of us much of the time).
3) Then, the other interface tweak (addressing this as well as another endless thread) would be to allow a preference to be set allowing a file to be opened and closed without saving, as long as NO OTHER CHANGES were made OTHER THAN assigning a new profile or ignoring an embedded profile, leaving the file in the same state it was before opening.
Both of these preferences should be set from within the Color Settings dialog, but could be augmented by as "Set As Default" button.
___________________________________________________
RICK GORDON
EMERALD VALLEY GRAPHICS AND CONSULTING
___________________________________________________EMAIL: rick@rickgordon.com
WWW: http://www.shelterpub.com
Date: Tue, 21 May 2002 09:43:29 -0400
From: Scott Olswold
Subject: RE: EXIF, sRGB, and Embedded Profiles
Andrew,I see what you're moving towards...it's very similar to the older Deke McLelland tutorial on the Photoshop page, where he talks about the benefits of the embedded profile by way of the display correction. You're actually saying that in conditions 2 and 3, you're trying to make it look good because you have some control in the matter. To some degree I do the same thing in Acrobat 5 every once in awhile using the Pitstop plugin. More often than not it fixes things (and more often than not, those images *are* tagged wrong).
Chris,
Thanks for the clarification...but I wonder if the average consumer really even notices? I've not bothered with a digital camera yet, the ones I like are out of my price range right now (yeah, yeah, it's a business expense, but some of my business expense comes out of my personal coffer, and the personal finance controller isn't up to spending on *another* camera just yet). A good friend has a CoolPix something that seems to do really well, at least in creating a nice, sharp photo, but the color is somewhat washed out. Sounds like an experiment for the weekend.
Scott Olswold
Date: Tue, 21 May 2002 21:25:47 +0300
From: "Yzzet"
Subject: FW: EXIF, sRGB, and Embedded Profiles
It seems to me that if the job on hand is NOT only an open & print job; i.e. you have to enrich the image all 1-2-3 below boils down to the same thing. In most instances you do not know if the profile embedded is right, it is also immaterial, you have to check and see if the image on your calibrated monitor is what you want (pleasant, more communicative, whatever).If the image is not going to be printed in a shop, there is another problem; it will most probably be viewed on uncalibrated media (I do not think the masses will ever calibrate their monitors, view the images on the monitors in controlled light situations, etc.), it may be printed on different inkjet printer-paper-ink combinations. In this situation there will be millions of output spaces which are undefined.
I wonder how much of this talk about color management, profiles, etc. is hairsplitting, and how much of this has practical value for the masses.
There may be some museum-like viewing places in the future where connosieurs of art will go to see the real intent of the creator of the image (like going to a concert hall vs. hearing to the same tune on the car radio while driving). Some people may form hi-fi imaging clubs to gather and evaluate monitors, monitor cables, etc.
Sorry I get carried away..
Izzet Cigirgan
Date: Tue, 21 May 2002 11:26:01 -0600
From: Chris Murphy
Subject: Re: EXIF, sRGB, and Embedded Profiles
John writes:> I use a large commerical color lab for most of my finishing. They have
>always said to use sRGB.
>Since I have started reading books and following these discussions on line,
>I hear a lot of people say not to use it. I use a Hasselblad with color
>negatives and then if I need digital work, I either have the lab do all of
>the digital retouching or I have them scan my negative and then I do the
>retouching in Photoshop and everything seems to work fine.>I'm still confused as to why this is not the best way.
If your stuff is coming out the way you expect it to, I wouldn't worry about it. For offset printing, I'm not that concerned about sRGB being used as source because in the real world it doesn't usually matter. Sometimes it can, but usually it doesn't. For a photo lab on the other hand, making prints let alone transparencies from a film recorder, I don't think sRGB is ideal. You will still get what you're expecting if you us sRGB, so it's not like it's a disaster. But something that can output prints (like a Lightjet, or Pictrography) or a film recorder has a gamut much larger than sRGB. If you use sRGB, the colors that exist with those output processes will never get used. So that's the problem - you're not maximizing the capabilities of the output device. It's not like your images are going to get trashed if you use sRGB.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Wed, 22 May 2002 20:58:48 -0400
From: "Calvin Jones"
Subject: The sRGB-in-the-digicam-EXIF conundrum
Fact: Digital cameras of all levels of cost/quality have inserted a value into an EXIF field claiming the color values in their image files should be interpreted in the sRGB color space.Fact: A small number of the higher-end cameras are actually using a different color space to write values - Adobe RGB - when the user has selected that space.
Fact: PS 7 is confused by that EXIF datum, and makes life a mess for many.
If one assumes either that there are NO capabilities in the digicam to select a color space of any name, or if it IS possible to select, that the user HAS selected sRGB, then the field in the EXIF may or may not be "importantly" wrong.
My questions for the experts of this formidable list are:
1. If we ignore the high-end cameras that can write image data files to the Adobe RGB space and focus only on all others, are there any data - not theories of likely results, but actual data - that demonstrate that loading the image file into PS7 "accepting" the sRGB profile assumption will yield inaccurate results. Is anyone looking at the extent to which digicams from different manufacturers produce image files that conform or depart from the sRGB space?
2. Ignoring the answer to Q #1, what should we make of the claims about sample variations in CCD or CMOS chips (and for that matter any support chips) in digicams with respect to their ability to deliver consistent color values across individual units of the same brand and model?
Question 2 is actually more interesting to me - in the same vein as the issue of 8- vs 16-bit editing - in the sense that many individuals and firms are now attempting to convince us that individual camera bodies will have distinct, individual color recording patterns that - if not profiled for the individual camera (not just the camera model) - will leave us with inaccuracies that we shouldn't tolerate.
It sounds quite plausible. But is there a body of evidence - derived from testing reasonable numbers of samples of the same camera model - that demonstrates conclusively that sample variations are larger than model or manufacturer variations - or at least greater than the threshold of human detection?
Like most photographers, I own one or at most two cameras of the same model. I can't demonstrate any significant differences. But what evidence is there that a profile for my unique camera serial number will - used properly - yield noticeably more accurate results than either a profile for a specific camera model, or even (shudder) either the generic sRGB or Adobe RGB profiles alone?
Thanks for your consideration and wisdom...
Calvin Jones www.CCJonesPhotos.com
Date: Thu, 23 May 2002 06:42:19 -0600
From: Andrew Rodney
Subject: Re: The sRGB-in-the-digicam-EXIF conundrum
on 5/22/02 6:58 PM, Calvin Jones wrote:> Fact: Digital cameras of all levels of cost/quality have inserted a
> value into an EXIF field claiming the color values in their image files
> should be interpreted in the sRGB color space.
Some do, some don't.> Fact: A small number of the higher-end cameras are actually using a
> different color space to write values - Adobe RGB - when the user has
> selected that space.
Lots of high end cameras shoot into something nothing like sRGB OR Adobe RGB.
> Fact: PS 7 is confused by that EXIF datum, and makes life a mess for many.
It1s confused because someone else is writing incorrect data...
>
> 1. If we ignore the high-end cameras that can write image data files to
> the Adobe RGB space and focus only on all others, are there any data -
> not theories of likely results, but actual data - that demonstrate that
> loading the image file into PS7 "accepting" the sRGB profile assumption
> will yield inaccurate results. Is anyone looking at the extent to which
> digicams from different manufacturers produce image files that conform
> or depart from the sRGB space?
>
Look at the data on your calibrated screen with the data tagged as sRGB..> 2. Ignoring the answer to Q #1, what should we make of the claims about
> sample variations in CCD or CMOS chips (and for that matter any support
> chips) in digicams with respect to their ability to deliver consistent
> color values across individual units of the same brand and model?
They are all different. That1s why we need custom ICC profiles for each camera, even from the same make and model.
>
> Question 2 is actually more interesting to me - in the same vein as the
> issue of 8- vs 16-bit editing - in the sense that many individuals and
> firms are now attempting to convince us that individual camera bodies
> will have distinct, individual color recording patterns that - if not
> profiled for the individual camera (not just the camera model) - will
> leave us with inaccuracies that we shouldn't tolerate.No, you can tolerate them and for many users, the output from the data will be fine. It can be better...
Andrew Rodney
Date: Thu, 23 May 2002 15:08:52 -0600
From: Ross Fenmore
Subject: Re: The sRGB-in-the-digicam-EXIF conundrum
Why can't digicam manufacturers create a generic profile for a specific camera model, and then convert internally to a common RGB space with a large enough gamut to encompass all possible colors? That would give the user who doesn't care to profile their camera a starting point to work from.If manufacturers chose sRGB, how bad would that be? Would sRGB be a generic space best suited for the typical PC user who may not even use a color managed system?
I've never profiled a digital camera. How well does it work? Do you need multiple profiles for different color balance and exposure settings?
If a profile has no useful meaning, then I think one should not be embedded or "announced" via EXIF. If I have to search for a profile to assign, then I'd just as soon start with an untagged file. Tagging a file with an irrelevant profile just creates confusion and compromises Color Management in general.
As far as Photoshop honoring the sRGB setting in EXIF data--I pay Adobe to make my life easier, not force me to deal with mistakes made by camera manufacturers.
Ross
Date: Fri, 24 May 2002 02:55:19 -0000
From: "wrandyr"
Subject: Re: The sRGB-in-the-digicam-EXIF conundrum
--- In colortheory@y..., "Calvin Jones"wrote:
> 1. If we ignore the high-end cameras that can write image data files to
> the Adobe RGB space and focus only on all others, are there any data -
> not theories of likely results, but actual data - that demonstrate that
> loading the image file into PS7 "accepting" the sRGB profile assumption
> will yield inaccurate results. Is anyone looking at the extent to which
> digicams from different manufacturers produce image files that conform
> or depart from the sRGB space?Since we are ignoring high-end cameras, my experience as a photofinisher printing files from various cameras indicates that there are more gross errors in the color accuracy than those caused by whether or not a suitable profile is or isn't there. Color balance is very inconsistent, in ways that cannot be explained by general lighting variations. Also, the color balance typically shifts when the flash is used. One would expect that with auto white balance the color coming off of digicams would be balanced better than with film, but so far that hasn't proven to be the case.
Randy Wright
Color Services
Date: Sat, 25 May 2002 04:58:16 -0500
From: Ron Bean
Subject: Re: The sRGB-in-the-digicam-EXIF conundrum
Randy Wright writes:>Since we are ignoring high-end cameras, my experience as a
>photofinisher printing files from various cameras indicates that
>there are more gross errors in the color accuracy than those caused
>by whether or not a suitable profile is or isn't there. Color balance
>is very inconsistent, in ways that cannot be explained by general
>lighting variations. Also, the color balance typically shifts when
>the flash is used. One would expect that with auto white balance the
>color coming off of digicams would be balanced better than with film,
>but so far that hasn't proven to be the case.From the reviews I've seen (www.dpreview.com, etc), no digital camera has an auto-white-balance that works properly in anything other than daylight. The presets (typically sunny, cloudy, incandescent, flourescent) are a little better, but they really only get proper color with the "custom" setting (point the camera at a white card and push a button). Flash is a problem because it's not the same color as the ambient light.
Of course it's not going to be easy to convince a casual photographer to reset the white balance every time the lighting changes. Some video cameras reflect ambient light off an internal white reference to set the white balance, but I don't know of any digital still cameras that do that-- as far as I know, they're trying to guess based on the image data alone.
Date: Sun, 26 May 2002 09:08:23 -0700
From: Jan Steinman
Subject: Re: The sRGB-in-the-digicam-EXIF conundrum
>The presets (typically sunny, cloudy, >incandescent, flourescent) are a little better...The Olympus E-series cameras (at least) have presets calibrated in degrees Kelvin. This makes it simple to use with a pro color meter. (The custom setting goes into the "0 degrees" slot -- too bad they didn't see fit to give you a read-out. Then the camera would BE a color meter...)
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Jan Steinman writes:>The Olympus E-series cameras (at least) have presets calibrated in degrees
>Kelvin. This makes it simple to use with a pro color meter. (The custom setting
>goes into the "0 degrees" slot -- too bad they didn't see fit to give you a
>read-out. Then the camera would BE a color meter...)How does this work with fluorescent lights?
Degrees Kelvin allows you to shift "toward red" or "toward blue", but flourescents have too much green, so you need to go "toward magenta". How many degrees Kelvin is that?
Fuji has a film with a 4th layer that's sensitive to the extra green in fluorescents, and effectively filters it out. I'm surprised nobody has done this with a CCD yet.
Date: Tue, 28 May 2002 13:43:02 -0700
From: Jan Steinman
Subject: Re: The sRGB-in-the-digicam-EXIF conundrum
> From: Ron Bean
>How does this work with fluorescent lights?
Point taken. "Degrees Kelvin" is only useful for Gaussian distributions, which florescent lighting generally is not.
I don't have the manual in front of me, but I seem to remember that two of the so-called color temperature settings were specifically for florescent light, and supposedly have a "spikey" response to complement that of florescent light.
I haven't shot under such light, so I can't say how well it works, but it sounds sort of like what you were writing about:
>Fuji has a film with a 4th layer that's sensitive to the extra
>green in fluorescents, and effectively filters it out.
>I'm surprised nobody has done this with a CCD yet.--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Wed, 29 May 2002 10:13:11 -0500
From: Ron Bean
Subject: Re: The sRGB-in-the-digicam-EXIF conundrum
Jan Steinman writes:>I don't have the manual in front of me, but I seem to remember that two of the
>so-called color temperature settings were specifically for fluorescent light,
>and supposedly have a "spikey" response to complement that of fluorescent light.
>I haven't shot under such light, so I can't say how well it works, but it sounds
>sort of like what you were writing about:>>Fuji has a film with a 4th layer that's sensitive to the extra
>>green in fluorescents, and effectively filters it out.
>>I'm surprised nobody has done this with a CCD yet.But that's just software, right? I was thinking about a CCD with a 4 color filter mosaic rather than 3 colors-- instead of G-R-G-B, it would be G1-R-G2-B. This would allow it to auto-detect fluorescent lights and respond intelligently, but I don't know if it would degrade the camera's performance under normal lighting.
Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.