Photography is NOT Reprography!

[eronald]

I have put up a new blog article describing my views on camera profiling. Enjoy !


Photography is NOT Reprography!

Edmund

[Schewe]

I have put up a new blog article describing my views on camera profiling. Enjoy !
Photography is NOT Reprography!

Edmund
[a href=\"index.php?act=findpost&pid=120820\"][{POST_SNAPBACK}][/a]

Nope, sorry ya still got it wrong. The two profile thing is correct (which is why Thomas DOES use two "profiles" in Camera Raw) but the reasoning you give is faulty because you are assuming the the sensor will respond the same to two different spectral sources which it doesn't.

The reason Thomas chose D65 & Standard Illuminate A was that sensors suffer from metameric failure. Meaning the spectral response at D65 will be different than the response under tungsten light. So no single profile will ever be able to describe that spectral response since it varies...

...and the tweening between those "profiles" is what makes Camera Raw capable of dealing with the adjustment of White Balance. And it's also the reason why running a calibration on a camera needs to be done at both illuminations-D65 & Tungsten.

[eronald]

Jeff,

 Please forward my link to Thomas -
 
 Thank you very much.

 Actually, in an ACR scenario, I would expect the "calibration profile" to be whatever ACR interpolates the color matrix for the set color temp. But there is no provision in ACR for previewing an adjustment layer.
 
Edmund

Nope, sorry ya still got it wrong. The two profile thing is correct (which is why Thomas DOES use two "profiles" in Camera Raw) but the reasoning you give is faulty because you are assuming the the sensor will respond the same to two different spectral sources which it doesn't.

The reason Thomas chose D65 & Standard Illuminate A was that sensors suffer from metameric failure. Meaning the spectral response at D65 will be different than the response under tungsten light. So no single profile will ever be able to describe that spectral response since it varies...

...and the tweening between those "profiles" is what makes Camera Raw capable of dealing with the adjustment of White Balance. And it's also the reason why running a calibration on a camera needs to be done at both illuminations-D65 & Tungsten.
[a href=\"index.php?act=findpost&pid=120842\"][{POST_SNAPBACK}][/a]

[bjanes]

I have put up a new blog article describing my views on camera profiling. Enjoy !
Photography is NOT Reprography!

Edmund
[a href=\"index.php?act=findpost&pid=120820\"][{POST_SNAPBACK}][/a]

Edmund,

As Andrew Rodney likes to point out, most photographers want pleasing color, not necessarily accurate color. You may remember from film days that Kodak strove to reproduce colors accurately with Ektachrome, whereas Fuji strove to get saturated colors with Velvia. Both films had their uses. Velvia was great for landscapes, but often did violence with Caucasian skin tones.

I see that you have a PhD, but do not state the field in which the degree was granted. Scientists and physicians use photography extensively to document their observations, and accurate reproduction of color is usually desired in this situation. Also, photographers shooting clothing or other merchandise for a catalog want accurate color, or else customers may return the merchandise because if it does not meet their expectations.

The eye can adapt to misrepresentation of many colors, but is intolerant to poor reproduction of memory colors such as human skin, green foliage, and blue sky. It will accept overly bright red flowers, but not a person who appears severely sunburned or jaundiced.

Bill

[eronald]

Dear Bill,

 I don't usually ask my friends about their studies, particularly as the best have often had rather strange trajectories. However, I'm quite willing to show off my PhD  if you want to come and see it and buy me lunch afterwards. The paper says "Applied Mathematics" and a line or so later "Algorithmics", whatever that means. I also happen to have an undergraduate maths degree,  and an engineering graduate diploma, and learnt to count when I was a kid and can still do that. If you ever come to visit, I'll be delighted to trade a print with you.

 A more interesting question is whether I have standing in the science community.
Maybe a perusal of the following link may convince you that some people have read, and discussed,  what I writ when younger ? Google search on one of my papers

Now, I believe that I've paid for my place at the table. Let me speak a bit about photography. As you stress, there is a place for "accurate" color, and a place for "pleasing" color, and then there are all those psychophysical complications which make the subject of color so difficult and interesting.

 In my blog proposal, I propose resolving the conflict between "accurate" and "pleasing" color by associating two things which *together* would create the color during a Raw file conversion:

- An "Calibration Profile" that would be established for every camera by whatever method the color calibration community agrees appropriate. ICC guys might choose an ICC input profile, DNG guys might choose their usual matrices with color temperature interpolation.

- Then a "Digital Emulsion" would be overlaid on top of the accurate color to create a pleasing "look". This could again be effected in a variety of ways - the ICC guys could use an ICC abstract profile, the Adobe guy could overlay an adjustment layer over the image while it's being twiddled for Raw conversion.

The main point above is that both communities, those who want accurate color and those who want pleasing color could get what they want: Someone who wants saturation could use a Velvia-type "Digital Emulsion" overlaid over the calibration profile of his camera. Someone who wants to do product would choose a more neutral rendering with the appropriate tone curve for product, again overlaid over the calibration for the camera actually used.

As far as workflow goes, the difference between my approach and what is done at present is that at the moment both the "Calibration profile" and the "Digital Emulsion" are collapsed into one profile, and I propose separating them clearly.

At present, if you use a "neutral" profile in the Raw converter, you cannot preview a look. So you move your files into Tiff, and then apply the look, but at that point the Raw conversion is already frozen.

This is because a converter only shows the effect of applying *one* profile at a time. I propose that it should be able to overlay two  - to paraphrase what my friend Guy Mancuso says, "as a Photographer you need to be able to see what you are doing".

Edmund

Edmund,

As Andrew Rodney likes to point out, most photographers want pleasing color, not necessarily accurate color. You may remember from film days that Kodak strove to reproduce colors accurately with Ektachrome, whereas Fuji strove to get saturated colors with Velvia. Both films had their uses. Velvia was great for landscapes, but often did violence with Caucasian skin tones.

I see that you have a PhD, but do not state the field in which the degree was granted. Scientists and physicians use photography extensively to document their observations, and accurate reproduction of color is usually desired in this situation. Also, photographers shooting clothing or other merchandise for a catalog want accurate color, or else customers may return the merchandise because if it does not meet their expectations.

The eye can adapt to misrepresentation of many colors, but is intolerant to poor reproduction of memory colors such as human skin, green foliage, and blue sky. It will accept overly bright red flowers, but not a person who appears severely sunburned or jaundiced.

Bill
[a href=\"index.php?act=findpost&pid=120852\"][{POST_SNAPBACK}][/a]

[PeterLange]

Please forward my link to Thomas -
Thank you very much.

Oh, please join the queue...
http://www.adobeforums.com/cgi-bin/webx/.3bc28bc5

  , Peter

--

[bjanes]

Nope, sorry ya still got it wrong. The two profile thing is correct (which is why Thomas DOES use two "profiles" in Camera Raw) but the reasoning you give is faulty because you are assuming the the sensor will respond the same to two different spectral sources which it doesn't.

The reason Thomas chose D65 & Standard Illuminate A was that sensors suffer from metameric failure. Meaning the spectral response at D65 will be different than the response under tungsten light. So no single profile will ever be able to describe that spectral response since it varies...

...and the tweening between those "profiles" is what makes Camera Raw capable of dealing with the adjustment of White Balance. And it's also the reason why running a calibration on a camera needs to be done at both illuminations-D65 & Tungsten.
[a href=\"index.php?act=findpost&pid=120842\"][{POST_SNAPBACK}][/a]

Jeff,

After hastily responding to Edmund's post without having studied the treatise on his web site, I actually studied that post and discovered that his suggestion was not quite what I had originally thought, and from the tenor of your post, I suspect that you may be in the same boat.

Edmund is proposing two profiles. The first is to obtain colorimetrically correct data and the second would then be applied to achieve an artistic intent. I don't see any inconsistency in his reasoning and Peter Lange (who knows quite a bit about color) has made similar suggestions.

Edmund's two profile suggestion has nothing to do with the two profiles used by ACR to achieve white balance. In this case, Thomas is trying to achieve colorimetrically correct data for white balance with multiple illuminants. Exactly how did Edmund get it wrong?

Bill

[Schewe]

Exactly how did Edmund get it wrong?
[a href=\"index.php?act=findpost&pid=120913\"][{POST_SNAPBACK}][/a]

Cause by his logic, we would need at least 4 profiles, one each for colometricly correct color at daylight and tungsten and again the two for creative rendering...

I really only care about two realities...my camera's capture of the scene and my rendered version of the scene-personally, I almost don't care what the capture looks like as long as it's reasonably well exposed and sharp. I can pretty much make anything look the way I want.

Now, if I were going to be pressed to have to match product color for a client for example, I might take a different approach but I've had a lot of experience in my earlier years trying to get accurate color on 8x10 chromes. I used to shoot for Budweiser and they were particularly fond of their color red. They always wanted it to be an "accurate" match to what they thought the color should be-by the numbers. Only problem is that if one did an absolute match to their PMS numbers, the shots always sucked cause everything in the shot was dictated by trying to get accurate numbers. I finally got fed up and told them, "Look, you can have accurate or you can have good-you want good, then we don't give a hoot what the numbers say, ok?". The work got a lot better after that...

There may indeed be situations where accurate is good for technical reasons. I know Bruce was consulting with the National Gallery in DC on their attempts at doing high rez captures of all their artwork. They wanted to match the paintings at all costs...and they were getting really, really close till somebody mentioned they could also match the paintings the way they were when the artist painted them as opposed to the way they are now after many years of aging...guess what? Yeah, you got it-nobody could figure out what to do....match the way the were when painted or match the way they were now after aging....as far as I know, they're still deciding...

But the whole issue about trying to achieve ICC profiles for digital cameras is the real problem. It's kinda hard to create profiles for a device that is subject to change based on the light under which it's used, the exposure and the method of processing. The whole idea behind making profiles is to profile a known and repeatable state...the moment you wank on raw processing controls you move off that known and repeated state. Bruce pretty much thought it was a fool's errand because he said digital cameras didn't have a fixed gamut of color, they had a color mixing function. So I see ICC profiles that are currently used in some raw processing software as just another color adjustment tool but one without sliders and knobs...and I like sliders and knobs because then I can control the results of the rendering. And I point out what I used to tell clients...you can have it good or accurate...I prefer good.

[digitaldog]

I have put up a new blog article describing my views on camera profiling. Enjoy !
Photography is NOT Reprography!

Edmund
[{POST_SNAPBACK}][/a]

[a href=\"http://www.color.org/ICC_white_paper_20_Digital_photography_color_management_basics.pdf]http://www.color.org/ICC_white_paper_20_Di...ment_basics.pdf[/url]

Note the site you're reading this white paper on as well!

First thing you must do is define what accurate color is (scene referred), then ask yourself if this is what you can (and want) to express output referred. Anything less isn't accurate. The term is used incorrectly by most color management manufacturers and others.

An accurate scan should reproduce the film as it appeared but its already output referred. Raw isn't, its scene referred.

As for 'look profiles' that in my mind is moving a rendering slider in a converter and saving it as a preset. I think the term look profile is basically misleading as most probably think of this as an ICC profile and we know that you don't need to use ICC profiles in all phases of work, as is seen in ACR (the two proprietary profile approach Thomas uses). Anything that alters the rendering you can save off and use on other images is a rendering preset.

[digitaldog]

Edmund is proposing two profiles. The first is to obtain colorimetrically correct data and the second would then be applied to achieve an artistic intent. [a href=\"index.php?act=findpost&pid=120913\"][{POST_SNAPBACK}][/a]

Eric Walowit is a pretty clever color scientist (he started Color Savvy, has a patent that is part of the core technology of ColorSync). He's on the ICC digital photo committee and has a concept that will probably work really well once the technology comes down in price AND (harder) the camera manufacturers get hip to his idea. In a nutshell, you build a small Spectrophotometer into the camera to measure the illuminant. That data is tagged to the capture. You have (somewhere) the spectral sensitivities of the chip. With that data, you could in theory, build on the fly, a 'profile' of each scene. You'd also use the EXIF data collected on exposure and capture to do all this.

Seems a lot smarter than using the camera with ONE illuminate and shooting a target with a fixed gamut, then trying to use that to describe every possible shooting condition we currently find ourselves in as the data to build an input profile. Treating a digital camera as if it were a scanner (the way the current technology works) is a huge stretch to be kind.

I would submit you don't need a profile to achieve an artistic intent. You need an artist (a human looking at the image output referred at the very least).

[eronald]

Jeff,

 I propose using two mechanisms - let's not call them profiles.

- The first mechanism is used to guess the "accurate" color, whatever that means. ACR for instance has the daylight-tungsten interpolation technique for this. This could very well be retained - why fix something that works?
- The second mechanism adds the "artistic" stuff. Photoshop can do this very well eg. with an adjustment layer -again, why fix something that works ? - but my innovation is that you would see the effect of your chosen artistic "look" eg. Velvia, while twiddling the exposure, brightness etc sliders of the Raw converter.

 In the ICC world, one way to obtain the same effect is to compose an abstract profile with an input profiles. There are many other ways to do the same.

 As you can see, Jeff, there is no *functional* novelty in what I propose, and thus no need for 4 profiles etc - if you have a working camera characterization technique which you like -and you seem to like the one in ACR- then you can extend it to my workflow.

 But the *creative* improvement in my proposed workflow is marked, because YOU CAN SEE WHAT YOU ARE DOING !

 As Bill suggested, above, you may have been somewhat hasty in reading my blog article, which is in fact quite spare, and too summarily written.

Edmund

Cause by his logic, we would need at least 4 profiles, one each for colometricly correct color at daylight and tungsten and again the two for creative rendering...

I really only care about two realities...my camera's capture of the scene and my rendered version of the scene-personally, I almost don't care what the capture looks like as long as it's reasonably well exposed and sharp. I can pretty much make anything look the way I want.

Now, if I were going to be pressed to have to match product color for a client for example, I might take a different approach but I've had a lot of experience in my earlier years trying to get accurate color on 8x10 chromes. I used to shoot for Budweiser and they were particularly fond of their color red. They always wanted it to be an "accurate" match to what they thought the color should be-by the numbers. Only problem is that if one did an absolute match to their PMS numbers, the shots always sucked cause everything in the shot was dictated by trying to get accurate numbers. I finally got fed up and told them, "Look, you can have accurate or you can have good-you want good, then we don't give a hoot what the numbers say, ok?". The work got a lot better after that...

There may indeed be situations where accurate is good for technical reasons. I know Bruce was consulting with the National Gallery in DC on their attempts at doing high rez captures of all their artwork. They wanted to match the paintings at all costs...and they were getting really, really close till somebody mentioned they could also match the paintings the way they were when the artist painted them as opposed to the way they are now after many years of aging...guess what? Yeah, you got it-nobody could figure out what to do....match the way the were when painted or match the way they were now after aging....as far as I know, they're still deciding...

But the whole issue about trying to achieve ICC profiles for digital cameras is the real problem. It's kinda hard to create profiles for a device that is subject to change based on the light under which it's used, the exposure and the method of processing. The whole idea behind making profiles is to profile a known and repeatable state...the moment you wank on raw processing controls you move off that known and repeated state. Bruce pretty much thought it was a fool's errand because he said digital cameras didn't have a fixed gamut of color, they had a color mixing function. So I see ICC profiles that are currently used in some raw processing software as just another color adjustment tool but one without sliders and knobs...and I like sliders and knobs because then I can control the results of the rendering. And I point out what I used to tell clients...you can have it good or accurate...I prefer good.
[a href=\"index.php?act=findpost&pid=120922\"][{POST_SNAPBACK}][/a]

[Schewe]

- The second mechanism adds the "artistic" stuff. Photoshop can do this very well eg. with an adjustment layer -again, why fix something that works ? - but my innovation is that you would see the effect of your chosen artistic "look" eg. Velvia, while twiddling the exposure, brightness etc sliders of the Raw converter.
[a href=\"index.php?act=findpost&pid=120939\"][{POST_SNAPBACK}][/a]

Uh huh. . .a preset...of predetermined creation and saved as a settings to achieve a "look".

Well, ok, Lightroom ships with some generic ones-Camera Raw doesn't and makes you create your own. And I suppose some photographers who choose not to learn how to use the controls in Camera Raw and Lightroom would opt for "generic looks" so they don't actually have to know what they are doing...somehow that doesn't appeal to me.

Each image I work on in Camera Raw or Lightroom pretty much tells me how to process it. Sometimes I may actually process an image twice or more because the current parametric controls are not sufficient and I have to do local adjustments...but a "look" to me is just somebody else's rendering interpretation and not my own which is a really foreign concept to me.

And for the love of god, don't make a "look" into anything even close to an "ICC Profile". Look at how screwed up the ICC is with regards to Perceptual mapping. If they can't get that right, what makes you want to hand over a concept of "artistic renderings" to something like the ICC?

And even then you have the problem of who decides "The Look"? Is there a "Look Master" who sits in judgement over what's a good look or bad look? Right now we see what's happening with Nikon & Canon and their "Looks". Their look is to make a capture look good in sRGB-to step on the shadow detail, punch up the color and hide their processing and lens defects in their processing to develop their "Look". Please don't tell me you want Nikon or Canon in charge of Look determination...

But the *creative* improvement in my proposed workflow is marked, because YOU CAN SEE WHAT YOU ARE DOING !

Well, I can see what I'm doing now pretty well, thank you...I also know what knobs to turn and what sliders to slide to get what I want. Perhaps that makes me unusual...I would hope not.

[englishm]

As much as I respect the thought you have put into this "two-profile" approach, I'm afraid that in in my opinion at least, the idea is bit constipated.  A number of thoughts come to mind, in no particular order:

I've shot thousands of rolls of various film emulsions over twenty-five years, and certainly have my favourites.  I resisted the switch to digital for a long while; unwilling to let go of what was familiar and predictable.  But once having made the switch, I'm not about to go back.  My point is that digital offers a number of new paradigms, and among them is the "digital look".  I see little point in trying to make digital behave like an old familiar film emulsion, no matter how much I liked it at the time.  It's a bit like wanting to make E100G look like Velvia pushed a stop.  And while I loved the saturated warm colours of Velvia, it had lousy skin tones and was often a bitch to scan, owing to its high Dmax and contrast level.  Digital has its own look, albeit one with a level of adjustment possibilities that film could never hope to achieve.

If you pine for the look of a particular film emulsion, then ACR/LR already offers an almost endless array of adjustment opportunities, which can be saved as a preset and applied at ingestion.  I suspect that it would be possible to measure the spectrometric response of film in much the same way as you can shoot a Gretag-McBeth 24 patch chart with your digital camera, and create a custom ACR calibration  together with tone curve and HSL settings, etc. to mimic the look of that emulsion; there might even be a business opportunity there developing and selling custom presets (as others already have in PS)... but I digress.

I'm not sure I understand what you mean when you point out that one of the benefits of your approach is that I could "see what I was doing"...  I can see quite clearly the effects of any changes I make in ACR/LR

Rather than being slavishly bound to the look of a particular film, I relish the creative freedom that ACR/LR and Photoshop allows me to interpret an image.  And to the extent that I can accomplish this "look" in ACR/LR, I accomplish this all in the wide-gamut gamma 1.0 colour space of the RAW file... applying the net result of all my cumulative adjustments only at the time of rendering the RAW file to .psd or .tif, or whatever.

At least to my way of thinking this appears to be a solution to a problem that either doesn't exist, or for which there would be limited appeal among many digital shooters.

Just my 2 cents.

[eronald]

Jeff,

 I don't have an issue with what you're saying here - in fact I pretty much agree with everything.

 I too create my own looks - and then ruthlessly reuse them over and over again in my own photographic work, thereby sell them to clients, and sometimes even sell them (relabelled as profiles) for C1 users.

 No, I don't want anyone especially "in charge" of look determination. I want pros to be able to mix up their own; I also want users to be able to purchase "canned" looks, digital emulsions,  from people like me or Andrew who know how to work with color but are not primarily photographers.

  As for seeing what you're doing at the moment, yes I'm afraid you are somewhat unusal. All master retouchers like us can work towards a target because they have an intimate mastery of the abilities of the tools, and their own vision tells them their goal. In this sense while developing your image in ACR you already know how you will retouch it with global edits in the main Photoshop program.

 This is not so for Joe User who needs to *SEE* what he is doing, as he does it. To him, knowing that a Photoshop action is available later for applying a "Velvia look" is not the same as seeing the Velvia look already applied while twiddling the Exposure and Brightness sliders in ACR.

Edmund

Uh huh. . .a preset...of predetermined creation and saved as a settings to achieve a "look".

Well, ok, Lightroom ships with some generic ones-Camera Raw doesn't and makes you create your own. And I suppose some photographers who choose not to learn how to use the controls in Camera Raw and Lightroom would opt for "generic looks" so they don't actually have to know what they are doing...somehow that doesn't appeal to me.

Each image I work on in Camera Raw or Lightroom pretty much tells me how to process it. Sometimes I may actually process an image twice or more because the current parametric controls are not sufficient and I have to do local adjustments...but a "look" to me is just somebody else's rendering interpretation and not my own which is a really foreign concept to me.

And for the love of god, don't make a "look" into anything even close to an "ICC Profile". Look at how screwed up the ICC is with regards to Perceptual mapping. If they can't get that right, what makes you want to hand over a concept of "artistic renderings" to something like the ICC?

And even then you have the problem of who decides "The Look"? Is there a "Look Master" who sits in judgement over what's a good look or bad look? Right now we see what's happening with Nikon & Canon and their "Looks". Their look is to make a capture look good in sRGB-to step on the shadow detail, punch up the color and hide their processing and lens defects in their processing to develop their "Look". Please don't tell me you want Nikon or Canon in charge of Look determination...
Well, I can see what I'm doing now pretty well, thank you...I also know what knobs to turn and what sliders to slide to get what I want. Perhaps that makes me unusual...I would hope not.
[a href=\"index.php?act=findpost&pid=120945\"][{POST_SNAPBACK}][/a]

[digitaldog]

On 6/3/07 3:28 PM, "Luminous Landscape Forum"  wrote:

> - The first mechanism is used to guess the "accurate" color, whatever that
> means. ACR for instance has the daylight-tungsten interpolation technique for
> this. This could very well be retained - why fix something that works.

You don't guess accurate color, you measure it (otherwise, how do you know it's accurate?). In ACR, when you correctly configure the calibrate tab, you're doing this by examining the numbers off a known target (Macbeth) that has been measured. This happens to be output referred but the numbers in the output referred color space matches the numbers you'd measure of the target, based on a certain output referred color space. Bruce did this in ProPhoto RGB.

> The second mechanism adds the "artistic" stuff.

Artistic stuff is the opposite. It has no actual basis in measurements, its up to the person rendering the image to interpret.

>  In the ICC world, one way to obtain the same effect is to compose an abstract
> profile with an input profiles. There are many other ways to do the same.

And who's using them? The ICC may have provided a spec for abstract profiles but by and large, they are a huge failure. Probably because its pretty easy and seamless to use an input or output profile in the workflow and when you want to alter the look, you use the myriad of rendering controls that software like Photoshop and raw converters provide. Building abstract profiles are not anywhere as easy to see and use by comparisons. The only functionality they might have had in the past was the ability to apply them to lots of images at once. Now that we have metadata editors, (Lightroom is a good example), this old technology seems even more doomed. Again, what's the difference in the end if you build and use an abstract profile or you build a set of rendering instructions? Ones easier to use. Now I'll grant you that a metadata edit in a converter like LR is proprietary to that product, an ICC profile isn't. But again, I really don't know anyone but a very few geeks who have used abstract profiles. BTW, have you seen the silly controls provided in the ColorSync utility for making them? Photoshop 1.0 was vastly superior. Of course, most of the functionality in the ColorSync utility is a joke.

[eronald]

Hehe Andrew, honest, I wasn't ignoring you, I was just figuring out a way to reply to your earlier post. Jeff is an easier sparring partner

 I'm well aware that you don't "guess" accurate color.  So I put in "guess" as an abbreviation for "use a method which Andrew would approve of". If you ask my opinion, I'm unhappy with setting in stone any method for going from scene-referred to output-referred. Anyway, I'd say that this part is for specialists.

 The artistic stuff is clearly for artists. Who need to be given nice artistic tools.

 Re. the abstract profiles, I don't insist they're the right way to do things, I just think that they might be one way to do it- a metadata editor would suit me just fine too, as would an adjustment layer in PS, it's just that I think the controls here should not be amalgamated with the Raw conversion settings, and also I insist on the previewing ability. Also, one nice thing about abstract profiles is that they're standardised, even if they're failures. Kodak used them btw.

 Deferring the preview of the "look" to after the conversion is leaving it too late.

 At the moment the tools in ACR or Lighroom look overloaded to me. The same sliders are used for everything: Individual image adjustments, camera color correction and "look".

 I'll have a look at the Colorsync tools for the abstract profiles ...

Edmund


 

On 6/3/07 3:28 PM, "Edmund M. A. Ronald"  did write:

> - The first mechanism is used to guess the "accurate" color, whatever that
> means. ACR for instance has the daylight-tungsten interpolation technique for
> this. This could very well be retained - why fix something that works.

You don't guess accurate color, you measure it (otherwise, how do you know it's accurate?). In ACR, when you correctly configure the calibrate tab, you're doing this by examining the numbers off a known target (Macbeth) that has been measured. This happens to be output referred but the numbers in the output referred color space matches the numbers you'd measure of the target, based on a certain output referred color space. Bruce did this in ProPhoto RGB.

> The second mechanism adds the "artistic" stuff.

Artistic stuff is the opposite. It has no actual basis in measurements, its up to the person rendering the image to interpret.

>  In the ICC world, one way to obtain the same effect is to compose an abstract
> profile with an input profiles. There are many other ways to do the same.

And who's using them? The ICC may have provided a spec for abstract profiles but by and large, they are a huge failure. Probably because its pretty easy and seamless to use an input or output profile in the workflow and when you want to alter the look, you use the myriad of rendering controls that software like Photoshop and raw converters provide. Building abstract profiles are not anywhere as easy to see and use by comparisons. The only functionality they might have had in the past was the ability to apply them to lots of images at once. Now that we have metadata editors, (Lightroom is a good example), this old technology seems even more doomed. Again, what's the difference in the end if you build and use an abstract profile or you build a set of rendering instructions? Ones easier to use. Now I'll grant you that a metadata edit in a converter like LR is proprietary to that product, an ICC profile isn't. But again, I really don't know anyone but a very few geeks who have used abstract profiles. BTW, have you seen the silly controls provided in the ColorSync utility for making them? Photoshop 1.0 was vastly superior. Of course, most of the functionality in the ColorSync utility is a joke.
[a href=\"index.php?act=findpost&pid=120952\"][{POST_SNAPBACK}][/a]

[Schewe]

Jeff is an easier sparring partner
[a href=\"index.php?act=findpost&pid=120958\"][{POST_SNAPBACK}][/a]

Ah, man...you shouldn't have said that bud. And here I was trying to be nice. Well, it's back to bare knuckles for me....

Ok, Edmund...what the heck are you trying to accomplish in all this. I mean, what's the point? To a degree, I would agree that "accurate" can be useful, sometimes, but that's rarely what photographers really want...they want to take reality and make it look neato. So, how ya gonna get "neato" out of a can?

You either know what you are doing, have an artistic point of view and the capacity and skill to do something or ya don't. Everybody wants to make a silk purse out of a sow's ear, but sometimes, you really only have a sow's ear, ya know? And some people just shouldn't be photographers-giving them a "neato button" ain't really doing them (and the industry) any favors.

The tools are getting better...CR 4.1 is better than 4.0 which is better than 3.7 which was better than 2.4, right?

So, you can either stand on the sidelines-as you do, and bleat about I don't know what-I don't know whether it's the fact you live in France or that you may have been dropped on your head as a child but sometimes it seems like stuff comes from your keyboard as recognizable words in random patterns-or get into the trenches and try to help the people that are at the forefront of what's happening.

So, do you want to help? If so, focus on something useful...

:~)

[eronald]

Jeff,

 I'm not sure you'd make more as a psychiatrist than as a photographer, but there might be hope for you as a lawyer when your eyes go bad  Let's hope that doesn't happen too soon.

 I'm outlining a software architecture for Raw which is real easy to implement, and which will make a lot of guys and gals happy.
 You're the exception, because you can already do everything you want real fast with the existing color tools - and so can I - but you're a top photographer working with Adobe, and I'm a guy who sells camera profiles, so we're not exactly typical of the average user. Here is a bestiary of who's really out there:

- Teddy Tourist wants a quick way to get standard "looks" for his landscape and city shots: Velvia style, nice contrasty monochrome etc. He just wants them quick and easy, as canned defaults, but he still needs to adjust exposure on the Raw because his metering is sloppy. Teddy also needs some quick soft pastel tones to impress his latest wannabe girl model.

- Pete Profiler works for a camera manufacturer and doesn't want to have to redo the looks Teddy expects for each camera , like I need to redo them at the moment for an unnamed client. Pete wants camera-independence at least for the look part so he can get his job done and the profiles out the door.

- Rita Assistant helps with the computer and does the touchup for older photographers who think Photoshop is something invented by Steven Spielberg. She is expected to retouch portrait sessions with her trademark "skin patina" for faces. But the photographers keep getting all these new cameras! So she needs a way to make the images look the same with every model of camera and every model of model.

- Frederic Fashion is a modern photographer who actually knows about retouching, he does his own, which keeps him at the screen in the evenings and cuts down on his social life. He cooks up a client look on a couple of files, now he wants to apply it quickly across a whole session, including some marginally exposed shots. Frederic wants one set of sliders for a look, different ones for the exposure settings, because he needs to be able to slap a new look on the files even after the exposure has been changed.

Now all the guys above benefit from the separation implied by my scheme. Calibrate the converter for the camera, then add on the look. Adjust the exposure of each shot, then apply the look. Calibration, individual image adjustment and look become independent components which each user can twiddle independently without wrecking the other parts.

I think it's a neat concept, more so since it builds on the existing toolkit.

Edmund

So, you can either stand on the sidelines-as you do, and bleat about I don't know what-I don't know whether it's the fact you live in France or that you may have been dropped on your head as a child but sometimes it seems like stuff comes from your keyboard as recognizable words in random patterns-or get into the trenches and try to help the people that are at the forefront of what's happening.

So, do you want to help? If so, focus on something useful...

:~)
[a href=\"index.php?act=findpost&pid=120983\"][{POST_SNAPBACK}][/a]

[digitaldog]

I'm unhappy with setting in stone any method for going from scene-referred to output-referred. Anyway, I'd say that this part is for specialists.

I don't know that anyone is setting this in stone. Scene referred is scene referred so that part is set in stone. Output referred is totally up to the person controlling this process. So I'm a bit confused about your point.

Re. the abstract profiles, I don't insist they're the right way to do things, I just think that they might be one way to do it- a metadata editor would suit me just fine too, as would an adjustment layer in PS, it's just that I think the controls here should not be amalgamated with the Raw conversion settings, and also I insist on the previewing ability.

Again, I'm not following you. You can see all the effects of rendering settings in say LR or ACR.

Also, one nice thing about abstract profiles is that they're standardised, even if they're failures. Kodak used them btw.

Its moot if no one uses them. Also, where would you use them instead of rendering the image n a converter? Yes, you could set some converters to produce scene referred, then you want to open them in say Photoshop and apply an abstract profile? Why not just finish the job in the converter or just use Photoshop now that you have a pixel based image? Personally I'd far prefer to do all the heavy lifting in LR. There are so many advantages to doing the global tone and color using metadata instructions. The only downside I see is that you have to use that converter since all the instructions are specific and proprietary to that converter. But just look at the history that remains with the raw file in LR. Can't do that in Photoshop!

 

Deferring the preview of the "look" to after the conversion is leaving it too late.

Again, I don't understand. The preview is the preview at any time with in the editing (rendering) process.

 

At the moment the tools in ACR or Lighroom look overloaded to me. The same sliders are used for everything: Individual image adjustments, camera color correction and "look".

Well that could be a UI issue for you (and others). But fundamentally I see no difference in sliders used for image adjustment and look. They are the same. In fact, one could argue that the calibrate tab is just another set of rendering tweaks. Everything is a rendering (look, adjustment) tweak. Just as in Photohsop, you don't have to use every adjustment tool provided, the same is true in LR. For some, a click on the Auto Tone and WB may be all they desire to produce that one rendering.

As for profiles, I think they should be kept simple and used to define device behavior, something they do reasonably well, with the exception of most DSLRs!