21 Replies Latest reply on Oct 29, 2008 3:53 PM by (Marco_Ugolini)

    Color Management Workflow

      I understand the theory behind how ICC based color management works but would like to better understand how ACR and Photoshop store and manipulate the color information as it executes various actions such as Mode>Lab or Mode>CYMK and how to minimize adverse effects of editing RGB files in CYMK or Lab space

      1) What format is the RAW file? How is it mapped from RAW to the monitor in ACR? What controls the rendering intent used?

      2) How is pixel color defined when opened in CS4 from ACR? RGB, Lab, other?

      3) If Adjustment>Mode is changed from RGB to the larger Lab space then back to RGB without making any other changes why does a 0.255.0 green not revert back to 0.255.0? I tried this with Absolute Colorimetric and Relative Colorimetric and both returned green slightly contaminated with small amounts of R and B. Calculation error? What are the downside risks for making an RGB > Lab > RGB mode change for editing? Since Lab is the reference space it would seem in theory there shouldn't be any.

      4) I realize that making changing Adjustment > Mode to CYMK utilizes the defined CYMK pre-set (e.g SWOP CYMK) and this will also greatly compress the RGB gamut if mode then is switched back to RGB. Using a artificial CYMK profile like WGCMYK, GCR, Light b85 11-14-01.icm which appears to have a shape identical to AdobeRGB seems to minimize the changes in a RGB > CYMK > RGB mode switch. What are the downside risks for making an RGB > CYMK > RGB mode change for editing?
        • 1. Re: Color Management Workflow
          John Danek Level 4
          I'll take a stab at some of your questions:

          1.) Not sure about RAW, only that your camera's capture will be in some form of RAW and may or may not establish an RGB profile. And, not sure what "ACR" refers to. However, Rendering Intent is usually established in PS's Color Settings.

          2.) I believe it will come in as RGB.

          3.) To LAB and back again will be adjusted per your profile and/or color settings. I prefer "Perceptual" Rendering Intent in Photoshop images.

          4.) Seldom would I ever go to CMYK from RGB and then back again to RGB. You are correct that the gamut will be clipped when going to CMYK, that's why it's critical to do most, if not all of your color adjustments in LAB and/or RGB and then finally converting a copy to CMYK, so keeping an original file intact in RGB is crucial. There may be a situation where you've got a good RGB in 16-bit ProPhotoRGB, you convert to CMYK ( which just clipped alot of gamut to 8-bit, and then back again to RGB...what RGB are you coming back to? Adobe RGB? From CMYK? Not sure I understand the logic.
          • 2. Re: Color Management Workflow
            Level 1
            have you reviewed

            Real World Camera Raw with Adobe Photoshop CS3 (Real World) (Paperback)
            by Bruce Fraser (Author), Jeff Schewe (Author)

            I recall Jeff said recently his new Camera Raw book for CS4 was due out soon
            http://photoshopnews.com/
            • 3. Re: Color Management Workflow
              Gernot Hoffmann Level 3
              Chuck,

              I've done some tests in PhS CS2:
              RGB=sRGB
              Dithering=Off
              Doc.mode: 8 bit per channel
              Green=0/255/0
              Lab=88/-79/81
              Conversion RGB-->Lab-->RGB
              Green=9/255/0
              Lab=88/-79/81
              Indeed an error of dR'=9 of 255, but this
              has to be interpreted in the linear space as
              dR=255*(9/255)^2.2 = 0.16 of 255
              Not nice but practically irrelevant.

              There is no indicated error for numbers 0..255 in
              Doc.mode: 16 bit per channel.

              By the way: The profile connnection space is CIE XYZ
              for RGB-RGB conversions and Lab if CMYK is involved.
              But Lab can be the editing space:

              Dan Margulis had shown in his book 'Photoshop Lab'
              that multiple conversions RGB-->Lab-->RGB don't do
              any harm. One reason, besides the sufficient numerical
              accuracy, is the fact that RGB values in a photo don't
              reflect an eternal truth but just the actual state of
              the image processing (which means always: change the
              numbers).

              Best regards --Gernot Hoffmann
              • 4. Re: Color Management Workflow
                Level 1
                Gernot,

                Your 8 bit test of RGB>Lab>RGB mirrored my results. I hadn't thought to try the same test in 16-bit mode and the fact there is no error confirms suspicions of it being calculation error.

                I have Dan Margulis' first book "Professional Photoshop" but haven't gotten around to buying the Lab one yet. Have you ever tried the wide gamut CYMK profile? I found it on the net here: http://www.curvemeister.com/tutorials/widegamutcmyk/index.htm
                but after re-reading Professional Photoshop recently also found that it was mentioned and included on the CD which came with the book. It appears in ColorSync to be same size and shape as AdobeRGB and doesn't clip the color much compared to when the device dependent CYMK gamuts such a SWOP CYMK are used.

                As I understand the ICC concept correctly, if Absolute Colorimetric intent is used and arbitrary RGB pixel values associated a smaller gamut (to calculate their position in Lab space) is converted to a larger gamut the absolute color definitions if the pixels in Lab space would not change, and neither would their arbitrary RGB values either because there would be no need to re-map the RGB: they fit in both spaces. Thus any error in a small gamut > big gamut conversion is likely calculation error due to estimating the outer range of the gamuts between known values in the CLUT ( color look-up table) for the profile. Correct assumption?

                So I think when people say a RGB>CYMK >RGB transform is destructive they've probably never tried it with a wide CYMK editing space selected in the color preference dialog. There is no reason conceptually that someone couldn't construct a CYMK profile for a hypothetical space the size and shape of CIE*Lab, which should then exhibit the same minimal gamut clipping as in a RGB > Lab > RGB transform. No?

                It seems some experts reject the concept of using a Wide Gamut CYMK space as being so impractical that its not worth even considering. I agree that is true, but at the same time not considering or using that possibility has led to the widely held belief that making any RGB > CYMK > RGB is akin to worshiping Satan; i.e. your color "integrity" will burn in Hell. Like Margulis I realize its all fake color anyway so if you can't see the flames in the output you shouldn't worry about the heat (i.e. minor calculation induced color errors). :)

                My questions are more out of curiosity than practical need. The only time I usually venture out of RGB nowadays is to create CMYK or Lab copies of a file to use Apply>Image to enhance detail in primary and secondary colors: a trick learned from the first Margulis book.

                I come from a background in CYMK printing in the early 1970s when direct camera separation and wet dot etching were still commonly used so I understand DM's "If you can't see it in the final output, it probably doesn't matter" color space agnostic viewpoint and early skepticism of the entire ICC based approach: It was touted as the Holy Grail which would allow everyone to see the same color, but even in our in-house facility where we could control and profile everything from designer workstation to offset presses it is still in 2008 more like herding cats than getting ducks in a row and quacking in unison. And that's with a $200 per hour color expert (not me) arranging the ducks.

                I used one of the first Cromalin pre-press proofing systems to hit the market in 1973, back when the toners were applied by hand: I worked in the National Geographic photomechanical lab at the time and we bought it for proofing map separations. I learned about the concept of 3D color space at a training class at DuPont from the color scientist who invented the Cromalin process. He used a plexiglass cube to represent Lab space with pins and colored yarn to show the Cromalin and press ink gamuts. Once I saw the 3D model the GATF hex plots of color space made sense. Duh! the hexagon represented a cube.

                The large web printer where I worked in the mid-late 1970s was part of the committee which developed the SWOP standards. In those days "Color Management" consisted of checking supplied 4/C separation film for ads against "progs" (single color proofs) with a 10x loupe to check dot structure and gain, and ink densities and trapping to ensure the color separator hadn't used an exotic set of high pigment inks for the proof we couldn't match on the web press. Color separators of Lipstick ads would sometimes use 5th color Rhodamine Red "kiss" plate on the proofs to please the "Match the Art" demands of clueless art directors, leaving the web printer "holding the bag" when the ad in the magazine didn't match the proof. The SWOP ink standards where primarily an attempt to eliminate such shenanigans, allowing us a reason to reject proofs made with inks which didn't match SWOP. Soon Cromalins and Matchprint proofs which didn't have all the press variables replaced progressive press proofs and for a time color matching was more predictable from transparency > scanner > Matchprint > Press... assuming the press was in good mechanical condition and not slurring dots and the ink and paper were a reasonably good match to SWOP.

                When digital replaced transparencies an important benchmark for offset printing was lost. It was no longer possible to have the original chrome on a light table or projector and the Matchprint made from the separations available while making color decisions for a running press. All the per-image color management control up to the time the image is printed on an offset press can't predict what will happen when two images with radically different colors run in-line on press, sharing the same ink controls. For example a photo with a big blue sky will make it very difficult to control more subtle blue tones in a small area of another photo on the same press sheet beneath it. It adds a whole host of variables not encountered on a ink jet printer.

                I joined US Information Agency as a printing / publication specialist in 1982 and became production manager of its large plant in the Philippines in 1983. We produced scans on a Hell DC300, which actually had a 1K digital "core" memory - wire grid w. magnetic donuts. We bought one of the first electronic proofing systems from DiaNipponScreen. It cost about $80,000 and used a high resolution B&W video camera to capture images of the CYMK separation films placed on light table, then electronically recreated a color image on a very expensive RGB reference monitor. It created a reasonable facsimile image but couldn't accurately display critical highlight detail or the CYMK gamut. So long before I ever saw Photoshop I understood the shortcomings of trying to pre-visualize and "soft" proof CYMK on a RGB monitor.

                We were also early implementers of DTP color and Photoshop on the production side in the early 1990s but our in-house customers still were using transparencies for most color. USIA had used 4/C magazines as an outreach tool since the early 1970s, but in 1995 the was decision made to kill the magazines and shift to web-based distribution of information. That fostered a shift to digital originals on the client side, but created a new problem on the printing side when it was realized that notwithstanding the wonders of the web people still actually liked to read magazines and books. Now instead of asking "Why didn't you match my transparencies" art directors and editors would ask "Why can't you match the color on my monitor" and the real fun began.... By that time I was away from the technical / reproduction side working on business systems and didn't actually use Photoshop much until I went digital for my personal photography in 2000 where I am quite happy with "pleasing color" and don't own a printer larger than 8.5 x 11. I'd love to still have all the tools to create and analyze color and actually control it, but all things considered I've got other better things to do :)
                • 5. Re: Color Management Workflow
                  Gernot Hoffmann Level 3
                  Chuck,

                  thanks for the colorful story of your life.
                  We could have met in the Philippines, where I had
                  been often since 1980.
                  http://www.fho-emden.de/~hoffmann/howww04d.html

                  Sometimes I'm a calibrationist:
                  http://www.fho-emden.de/~hoffmann/calib-02.pdf

                  But on the other hand I'm believing in this workflow:

                  Soft Proof by PhS:
                  http://www.fho-emden.de/~hoffmann/labproof15092008.pdf
                  (this doc requires special color management settings in Adobe
                  Reader or Acrobat, because all images are in AdobeRGB)

                  Inkjet Proofs by a qualified RIP:
                  The inkjet proofs are used as references for adjusting the press.
                  Otherwise the press operators wouldn't have the faintest idea
                  how the print should look.

                  The calibrationism ends at the press. IMO, one should use
                  AVERAGE process descriptions like SWOP or ISO Coated
                  or newer versions of these.

                  I don't know the reasons for a wide gamut CMYK working
                  space. IMO, one has to check the printability of RGB images
                  for a specified real world CMYK process.
                  And one has to counteract, if the result shouldn't be satisfying.
                  Perhaps somebody else can explain the benefits of wide gamut
                  CMYK.

                  Best regards --Gernot Hoffmann
                  • 6. Re: Color Management Workflow
                    Level 1
                    Color is like any other language, you tend to think in the one you learned first and use the most. Working in pre-press and printing it took a long time to think of color in anything besides CYMK dot percentages and how they reproduce color with offset ink gamut. So its often easier for me to pop an image into CYMK to make a selective color tweek (usually on a separate layer with a mask). The advantage of the artificial wide CYMK gamut is it allows doing that without any significant clipping or color shift when switched back to RGB.

                    You see that same "thinking in CYMK" bias in Margulis' Professional Photoshop book: most of the examples are done in CYMK which was confusing and off-putting to photographers who thought of color only in terms of numerical RGB values on a scale of 0-255, with CYMK as a second language: with an RGB workflow you really don't every see the CYMK or for that matter realize things like UCR are occurring to keep total dot% of the four colors under 280%

                    RAW and the ability to adjust blue-yellow WB and green-magenta tint in Adobe Camera Raw and smart-objects have eliminate the need to use CYMK for tweeking, but sometimes its still easier for me to think in terms of CYMK than RGB.

                    Ink jet proofs are only as good as the icc profiles. Matchprint had the advantage of being made from the same film which made the press plates. We'd make an 8-page form Matchprint as a guide to what was actually in the film. That baseline made it easier to see when things like in-line inking or trapping was the cause of color matching problems. Of course the march of progress has eliminated the film step so the entire process hinges on accurate press profiles which are difficult to obtain on offset presses due to all the mechanical variables. The success of the entire color workflow hinges on the profiling which adds a lot of what is really non-productive overhead to the process: a full-time calibrationist is needed to run a printing plant effectively. Hence the move in offset printing to a RGB PDF pre-press workflow where the press prints a target, creates an on the spot profile on the actual ink batch / paper being used then sends it to the RIP to control the digital imaging of the plates.

                    Actually a lot of things end at the press because any pressman I've every known has 101 different unorthodox solutions to press problems, most of which the guys in the pre-press department don't know about. I once had an ink trapping problem which was totally puzzling until it was discovered the pressman had been doctoring the ink with lubricating oil.
                    • 7. Re: Color Management Workflow
                      (Marco_Ugolini) Level 1
                      Chuck Gardner wrote:
                      >Your 8 bit test of RGB>Lab>RGB mirrored my results. I hadn't thought to try the same test in 16-bit mode and the fact there is no error confirms suspicions of it being calculation error.

                      It's not a calculation error, as much as a
                      b necessary
                      rounding due to the limited precision of 8-bit calculations. In 16 bits, the rounding is comparatively infinitesimal, trivial for all practical purposes.
                      >and doesn't clip the color much compared to when the device dependent CYMK gamuts such a SWOP CYMK are used.

                      That "wide-gamut CMYK" space is
                      i also
                      device-dependent. The fact that it's wider that US Web Coated (SWOP) v2 does not make it device-independent. Its CMYK numbers have no objectively inherent color meaning without a way to relate them to either XYZ or Lab (which is what an ICC profile does).
                      > Thus any error in a small gamut > big gamut conversion is likely calculation error due to estimating the outer range of the gamuts between known values in the CLUT ( color look-up table) for the profile. Correct assumption?

                      ICC conversions find the best available colorimetric match based on a few parameters. If the rendering intent is Relative Colorimetric when going from AdobeRGB to Lab, the source's white point is mapped to the destination's white point. Therefore, other colors will also have to be scaled to different degrees to make the white point fit.

                      By the way, AdobeRGB, sRGB, ProPhoto RGB and other standard working spaces do not have CLUTs, because they are
                      b matrix-based
                      rather than being based on lookup tables. That means that they are not capable of using a Perceptual rendering intent when going to or coming from the Profile Connection Space (XYZ or Lab). The Relative Colorimetric intent is the only one available -- and Absolute Colorimetric is equal to Relative Colorimetric plus a non-scaled white point.

                      Whether done using an RC or AC rendering intent, a round-trip conversion from RGB to Lab and back to the same RGB space will produce results that are
                      i colorimetrically very close
                      to the starting colors. The numbers will not match exactly, but the color appearance (in device-independent terms) will be very close.

                      ICC conversions are not about matching the numbers: they are about matching
                      b color appearance.
                      >It seems some experts reject the concept of using a Wide Gamut CYMK space as being so impractical that its not worth even considering.

                      I fail to see the logic that supports the use of a "wide-gamut CMYK" space", but bristles at the use of an RGB color space like ProPhoto RGB. It seems to me like a waste of energies to go through such contortions. You will have to exercise caution anyway, whether you use one or the other.
                      > Like Margulis I realize its all fake color anyway

                      What "fake color"? There are
                      i visible
                      wavelengths that we humans can see and non-visible wavelengths that we cannot see (but other creatures may be able to perceive). None of them is "fake". They are all very real wavelengths.
                      >[the entire ICC based approach] was touted as the Holy Grail which would allow everyone to see the same color, but even in our in-house facility where we could control and profile everything from designer workstation to offset presses it is still in 2008 more like herding cats than getting ducks in a row and quacking in unison. And that's with a $200 per hour color expert (not me) arranging the ducks.

                      The ducks (and the consultant, who is not an omnipotent god, after all) are helpless if the water keeps being being poisoned by those who would rather keep doing things a certain established way than try a new way. Examples abound of forward-thinking printers and/or separators that use ICC workflows successfully -- which suggests that the touted "failure of color management" is more likely the failure (or refusal fueled by hostility) on the part of individuals and companies
                      i to understand or apply
                      ICC color management. Without motivation and willingness to explore, results will not accrue.
                      • 8. Re: Color Management Workflow
                        Level 1
                        Chuck,

                        It seems that you might need to re-evaluate why you think you need to go to CMYK if your final output is RGB. There are plenty of reasons for taking a CMYK image to RGB or Lab, tweak it and re-separate, but precious few for the other way around. That you "think" better in CMYK isn't good enough. It isn't that hard to "think" in RGB and L*a*b, and RGB is a far sight more intuitive than the other two. The one reason that has been touted is to manipulate the black plate for extracting more shadow detail. Maybe. Maybe not. In comparison tests I've done, they're about the same. You might want to pick up Dan's L*a*b book and really read it. Dan's earlier books were sort of CMYK centric on the surface, but not really when you thought about them. He's really about all color spaces. It's up to you to use the one's best suited to your image and your output.
                        • 9. Re: Color Management Workflow
                          Level 1
                          Peter:

                          If you got the impression I regularly edit in CYMK it is not correct. So no need to build an argument against it using it or question the way I think. As for which color space is more intuitive? That is situational, depending on whether you are operating a digital camera or doing a color OK on a web offset press at the time: you'll get blank stares from a pressman if you tell him to raise the green channel 5 points. I've done both and I think equally well in RGB and CYMK, and Lab too...

                          BTW, I didn't say Dan was CYMK centric, just that most the examples in his first book "Professional Photoshop" are and that it might make it more difficult for someone who was RGB centric to relate to them. I'd followed his magazine articles for years and didn't have any problem relating to the concepts.

                          I've been told the "integrity" of my pixels are at risk by using CYMK as if the integrity of a pixel was important as the virginity of a teenage daughter when as I see it the real issue isn't switching to CYMK and back, but rather switching to a CYMK device-centric gamut like SWOP which too small to envelop the complete RGB gamut. In other words the "integrity" issue is related to relative gamut size/shape of original and target profiles not the fact the switch is made to CYMK. If absolute colorimetric rendering intent is used to switch from a smaller to larger gamut there is no change in the actual color value in Lab space, so the "integrity" stays intact, give or take some rounding error.

                          My questions were in the form of examples to explore and better understand the specific methods used by Photoshop to map color from one gamut to another than for any specific practical use; i.e. if RGB>Lab>RGB works, why can't RGB>CYMK>RGB?

                          To my way of thinking the rationale for having a large CYMK gamut available for editing is as valid as using a large RGB gamut like ProPhoto: so colors are not clipped when gamuts are changed in the workflow. Is switching modes back and forth an optimal workflow? No. Do people occasionally do it? Yes, otherwise it wouldn't be on the Photoshop menu bar next to Lab would it? So it would behoove one to learn how to do use the Photoshop functions with as little damage as possible. The way to do that with an RGB>CYMK>RGB round-trip appears to be using a CYMK profile which is larger than the RGB working space, instead of SWOP CYMK.

                          I'm not pro- or anti-color management. I spent most of my career doing process control of one form or another in pre-press and printing. Its not that it doesn't work, but rather that I find it ironic the evolution of technology has been driven the quest to save money, but the more it evolves the more expensive non-productive overhead is needed to make it work. Profit in any business comes from minimizing overhead, like those $200-per-hour color consultants.

                          As in the 1970s if I really want to know what really critical color will look like when printed I do what you probably do: make a test print :)
                          • 10. Re: Color Management Workflow
                            (Marco_Ugolini) Level 1
                            >To my way of thinking the rationale for having a large CYMK gamut available for editing is as valid as using a large RGB gamut like ProPhoto: so colors are not clipped when gamuts are changed in the workflow. [...] The way to do that with an RGB>CYMK>RGB round-trip appears to be using a CYMK profile which is larger than the RGB working space, instead of SWOP CYMK.

                            But...why? Just so that you can do Apply Image or some such in CMYK and then go back to RGB? If that is the reason, I would argue that whatever you want to do in CMYK, or very close to it, can be done without effecting round-trip conversions of any sort. It's a matter of finding new methodologies and workflows that suit our needs, without having to revert to the older ways. Take it as a challenge, not a limitation.

                            >I'm not pro- or anti-color management. I spent most of my career doing process control of one form or another in pre-press and printing. Its not that it doesn't work, but rather that I find it ironic the evolution of technology has been driven the quest to save money, but the more it evolves the more expensive non-productive overhead is needed to make it work. Profit in any business comes from minimizing overhead, like those $200-per-hour color consultants.

                            You're not anti-color-management, but still, it seems that you just can't help making small of it when the opportunity presents itself.

                            What "expensive non-productive overhead", by the way? ICC color management properly conceived and applied saves time and consumables, cuts down press makeready considerably, reduces dot gain (with soft-proofing and filmless CTP), and more. I'm sure you heard of GRACoL. What is "non-productive" about that?
                            • 11. Re: Color Management Workflow
                              Level 1
                              Apply Image is a good example....

                              It is possible to enhance detail in saturated primary colors by applying a color channel carrying detail in an RGB file to the L layer in an identical copy converted to Lab. The Lab file is then converted back to RGB.

                              But I have a photo of a Dogwood and find the Y channel in a CYMK has the best detail in the flowers of all the possible R,G,B, C, Y, M, K layers. So I save two copies of the original file:
                              Mode>Lab
                              Mode>CYMK

                              Note: I have selected the wide gamut CYMK profile in my color preferences, so my CYMK mode change creates no significant changes in the file.

                              Then I open the Lab copy and apply the Y channel from the CYMK copy to the L channel, then convert the Lab file back to RGB.

                              I actually did that just now and posted it at: http://super.nova.org/TP/Dogwood_RGB_Y.jpg

                              So in that case I didn't do a RGB>CYMK>RGB round trip, I just wanted to carry all the original color detail into the CYMK conversion so it wouldn't get clipped. The wide CYMK profile let me do that.

                              The round trip was converting the Lab copy I applied the Y channel to back to RGB.

                              The final result looks very close to how I perceived the Dogwood by eye when shooting it.
                              • 12. Re: Color Management Workflow
                                Level 1
                                > Its not that it doesn't work, but rather that I find it ironic the evolution of technology has been driven the quest to save money, but the more it evolves the more expensive non-productive overhead is needed to make it work. Profit in any business comes from minimizing overhead, like those $200-per-hour color consultants.

                                Chuck,

                                Your assessment of the cost/benefit of color management in the offset world is wrong. It sounds like you're basing your opinion on conversations around the water cooler as opposed to actually seeing it action in the real world.

                                I'd suggest that you get out and visit some printers (both sheetfed and web) who are using G7 methodology to calibrate their entire workflows. They don't do this as a hobby, they're in it to make money. And they've found that the $10k or so that they spend to get calibrated (including the "$200/hr consultant") saves them so much money in the long run that it's crazy not to do it. The savings add up quickly from fewer rounds of proofs (usually one), shorter make-ready, less waste, shorter runtime, and better quality overall (fuller gamut.) A printer I work with calculated that they made their investment back in three months. That's good business.

                                A lot has changed since 1970.
                                • 13. Re: Color Management Workflow
                                  Ramón G Castañeda Level 4
                                  Completely OFF TOPIC comment on
                                  > is like any other language, you tend to think in the one you learned first and use the most

                                  This discussion is for another venue, but, actually, thought precedes language. Thinking is an abstract process, and not until we have the thought do we reach for the convention, language.

                                  It would otherwise be impossible for those of us who are multilingual to function at all.
                                  • 15. Re: Color Management Workflow
                                    Level 1
                                    Marco -

                                    I like that curve in multiply. Very nice.
                                    • 16. Re: Color Management Workflow
                                      (Marco_Ugolini) Level 1
                                      Thank you, Rick. It's one good way to bring out detail in "washed out" highlight areas.
                                      • 17. Re: Color Management Workflow
                                        Level 1
                                        Proving there is yet another way to skin a cat. This version took all of ten seconds in RGB. Highlight/Shadow filter, then Fade to Multiply and back off opacity to suit. The first thing I notice is not only the highlight improvement but the added bonus of better shadow contrast in the leaves.

                                        http://www.pixentral.com/show.php?picture=1YmnXYOEPP4JZKgOQbE7XIyHARQQo0

                                        Chuck - there are many ways to work on images, very few of them that much better than any one of the rest, but I would say that there is a lot to learn here. I've been in quite a few press rooms in the last twenty years or so, and the one thing that seems to be common among the pressmen was a strong resistance to this new fangled color technology and prepress. I can't tell you how many times I've been told that this crap does not and can not work, only to have the same people come up to me after the press check, where the final output was closer to my own calibrated Epson proofs than their own proofer, and ask me just how I was able to do what they've spent years chasing unsuccessfully.

                                        My experience closely mirrors what Rick is saying. A lot HAS changed in the last fifteen years or so.
                                        • 18. Re: Color Management Workflow
                                          (Marco_Ugolini) Level 1
                                          I agree with Peter's obervations.

                                          Whether color management's opponents realize it or not, the future will be color-managed. It may not be ICC color management exactly in the form that we know now, but it's still going to produce repeatable and predictable results, more cheaply than any of the alternatives, and with results that are far better than the competition.

                                          The opposition of the "old guard" only delays the inevitable. It just makes too much sense both quality- and money-wise for CM not to win out in the end. Still, some of us will take their sweet time realizing that, and others just won't.
                                          • 19. Re: Color Management Workflow
                                            Level 1
                                            Sorry this has rendered itself into a pro- anti- discussion. That wasn't my intent.

                                            The point of starting the thread was to address the question of whether or not a RGB>CYMK>RGB round trip with a >>>> wide CMYK gamut <<<<<  would be any different than RGB>Lab>RGB in Photoshop. Based on the way icc color management works there shouldn't be any difference providing the hypothetical CYMK editing gamut is large enough to fit the RGB space. With Absolute Rendering intent the Lab values would remain the same and when CYMK reverted back to RGB the RGB values would revert back to similar ones, within the range of calculation error. I never did get a clear answer on that one, just a lot of well meaning lectures on the foolishness of doing such a thing and better alternatives.

                                            If I seem antagonistic its because I view color management with both the sophistication of someone who understands it and has seen it used it successfully in a large commercial printing environment and knows how it can work, but faces the same frustrations as the unschooled masses at home because I just can't justify the investment. The reason I came here for advice is because I do encourage others in photo fora to use it and wanted to make sure I fully understood the implications of things like mode conversions correctly. Don't worry your points are well taken...

                                            I've actually been encouraging people to use color management for quite a few years. Back in 2001 when I was invited to keynote the Philippine Graph Expo show (I was director of the State Dept. printing plant there) my theme was "Color Management - The wave of the future". I cut the ribbon with the daughter of the President of the Philippines who ran the science agency at the time:
                                            http://super.nova.org/PX/RibbonCut.jpg Here I am describing the CYMK gamut I use :)
                                            http://super.nova.org/PX/WideGamut.jpg I also taught a class on digital photography at the show which included an introduction to color management.

                                            So I do agree with you it is brilliant technology, but at the same time feel it will only be brilliant technology EVERYONE, including Joe Pointnshoot can use practically and effectively at the point where the Photoshop Page Set-up menu has a "profile printer" button which will automatically cause ANY printer to create a custom profile on the paper in the tray, then automatically send the profile back to Photoshop for a soft proof. That's the "easy" button needed to make color management all the way to the printer accessible and really useful to the masses for screen-to-print visualization. Most digital photographers probably don't even know soft proofing exists is or why they might want to use it. One of the more common questions on photo fora is still "How can I get my printer to match my monitor".

                                            Doctors mostly see sick people and I suspect that professional calibrationists mostly deal with people motivated to make color management work effectively across an enterprise. But poll 100 art directors, designers at random what percentage would know how to use soft proofing and have profiles for the presses / paper their work is printed on? I still deal with many designers who don't have a clue about the technical aspects of color and don't want to know: its not their thing. They want the printer to match their monitor too, and introducing color management into their workflow is more a exercise in lowering their expectations than anything else.

                                            I LOVE COLOR MANAGEMENT but I also share Joe's pain. I'm on the horns of a dilemma at home where I really don't print much and I didn't even own a color printer until a few years ago. No real need when you have a 24" Rainbow dye sub and 52" ink jet at work. I can't really justify the HP9180 I'd love to have at home (more a space issue than cost) so I've gotten by with my trusty HP7980 which worked fine with OSX 10.3 and PS7. An upgrade to 10.4 had a domino effect. Photoshop 7 wasn't supported requiring upgrading to CS3 (it was overdue anyway) only to discover CS3 didn't support Photoshop managed color on the HP7980 (the option turn off the printer's drivers disappeared from the Paper>Color menu) and CS4 doesn't either.

                                            Save the advice on printers, please. I'm patient and will wait for the $125 self-profiling printer which will wirelessly soft proof images on the LCD of my Canon when I press the print button on the back. That is what the print button there for isn't it? :)

                                            Again, thanks everyone: its been an interesting dialogue.

                                            Chuck
                                            • 20. Re: Color Management Workflow
                                              Level 1
                                              Chuck,

                                              In order to make profiles both usable and practical, compromises are made. Targets are made of from a few hundred to a couple thousand color patches that cover the gamut of the output device. It would be impossible to make a direct one for one profile with lookup tables for every conceivable color, so there is a fair amount of averaging and smoothing going on in the math. This also helps keep the profiles ( LUT based profiles) from getting too large. Matrix based profiles are only a few kb but are limited in their function - y'know - no perceptual intent. Every time you make a conversion, the lookup tables reference the Profile Connection Space - usually Lab or xyz - and compute the new pixel values. Because there are some many values that fall between actual measured points, they get averaged. Most of the time the results are fine, but there will always be certain colors that suffer from a double round trip.

                                              It's possible to build numerically more accurate profiles, but they tend to make prints that don't look as good to the eye. It's possible to engineer the software to maintain color purity in saturated colors but then other relationships falter. There will always be compromises in color math and usually they are biased toward making a pleasing image.
                                              • 21. Re: Color Management Workflow
                                                (Marco_Ugolini) Level 1
                                                Chuck Gardner wrote:
                                                >The point of starting the thread was to address the question of whether or not a RGB>CYMK>RGB round trip with a >>>> wide CMYK gamut <<<<< would be any different than RGB>Lab>RGB in Photoshop.

                                                Yes, but your reason for an RGB > CMYK > RGB round trip itself would be so that you can use the functions (Apply Image, etc.) that you believe you must use for your image editing. So, rather than the point you raise, the proper question in my mind would be: why is it worth the bother to do that type of round-trip conversion, when the editing can be done as effectively without leaving the RGB working space?
                                                >If I seem antagonistic its because I view color management with both the sophistication of someone who understands it and has seen it used it successfully in a large commercial printing environment and knows how it can work, but faces the same frustrations as the unschooled masses at home because I just can't justify the investment.

                                                Is that a good reason to feel antagonistic? I encounter that type of frustration every day in my line of work with people who do not understand or appreciate CM workflows, or never have any time, money or patience for what it can offer them (how does the saying go? "There's never any time or money to do it right, but there's always time and money to do it over." Silly, but true, and it rarely causes most people to rethink what they do: they just work
                                                i harder
                                                at how they already do what they do, instead of exploring other ways to do it
                                                i differently).
                                                Still, true as that is, I do not turn this frustrating situation into a reason to disparage the real benefits of CM. I can tell one from the other, and it would not make any sense.
                                                >That's the "easy" button needed to make color management all the way to the printer accessible and really useful to the masses for screen-to-print visualization.

                                                Nice thought, but that kind of result will be possible when either (a) the printers are built to perform with a high degree of stability and within a tight tolerance, so that a canned profile describes their behavior accurately; or (b) the printer has a built-in spectro that creates profiles which reflect the printer's current state, and the profile is then properly used by the software in the most idiot-proof way imaginable.

                                                Either scenario is unlikely in my view, because both quality control (option a) and added hardware (option b) add to the price of the printer, and we all know that the great majority of "Joe Pointnshoot" users
                                                i love
                                                to save the penny on their purchase. So that, when they see the printer with the spectro which costs "x" more than the one without, guess which one they'll buy?

                                                I keep hearing about this dream of "one-click, hassle-free, idiot-proof color management", and I don't see it happening any time soon. For a while longer still, we'll need to click more than one dialog box, we'll need to overcome hurdles, and we'll have to be anything but idiots, and study up instead before results materialize.
                                                >Doctors mostly see sick people and I suspect that professional calibrationists mostly deal with people motivated to make color management work effectively across an enterprise.

                                                Please, the word "calibrationist" is an insult of Dan's creation, enthusiastically shared by Mike Russell, and thrown around the way the word "liberal" is by Republicans, to smear, tar and feather opponents. If you use CM, you calibrate and profile, period. If that makes one a "calibrationist", does that mean that Dan & Co. would prefer to work with tools whose behavior is neither precise nor repeatable? Let's use respectful words that reflect our actual intelligence, please, and let's stop portraying any one of us as the most simpleminded of operators.