17 Replies Latest reply on Jan 20, 2010 1:26 PM by Peter DL

    Why curves in ACR acts so different in comparison on other programms?

    amam123

      Similar question I made in LR forum but realize that in here more suitable place.

      "Does anybody knows, why Adobe made adjustment of curves in lightroom so different in comparison of curves Photoshop? I mean not of adjustment method, but colour shifts. Darkening by curves in LR makes skin tones very dirty yellow-orange, while photoshop's tends to red, and when lightening in LR skin tones becomes _unpleasant_ pink-orange while in PS becomes acceptable uniform yellow hue. This I checked on the same file (tif). Differenses of colours shifts (LR vs PS) happens even while adjust other sliders, such as brightness and contrast, for example"

      As to my taste and preferences, I dont like LR behaviour.

      ps. In attachment you can see huge differences. Question is why? For what reason? And will I got this in future versions?

      pps. camera profiles and untwisted variants of them have not relation to this behaviour.

        • 1. Re: Why curves in ACR acts so different in comparison on other programms?
          Robert Shomler Level 4

          Photoshop’s curve operates on gamma-encoded image data; the LR (using Camera Raw engine) curve operates on the raw linear-gamma data.  You might find this PhotoshopNews article of interest:.

           

          http://photoshopnews.com/2005/04/06/bruces-curves/

          • 2. Re: Why curves in ACR acts so different in comparison on other programms?
            amam123 Level 1

            Thank you for this link.

            I have made next test: converted original image to ProphotoRGB space with gamma 1.0 and applied same tone curves in Photoshop.

            In result still have got not identical and far better colors in PS than in LR. Where my mistake?

            And opposite task: how to simulate LR-curves in PS? I want to understand this behaivour.

            • 3. Re: Why curves in ACR acts so different in comparison on other programms?
              Tim Lookingbill Level 1

              If you want to get an idea why there is this behavior you'ld have to talk to the programmers who wrote the code that controls the interplay between ACR/LR's tools, the numbers in the image and the video card.

               

              You'll probably need a week and then some to completely comprehend it because it involves color science, math and A/D converter electronics that controls voltage ratings to each pixel on screen and probably other stuff I don't care to know.

               

              But I can only suggest the following to give you an idea that what you are seeing is a gamma encoded environment influenced effect along with the complicated stuff mentioned above.

               

              Create in Photoshop a 1.0 gamma version of an oddball color space like CIERGB by going into Color Settings RGB Working Space menu selecting CIERGB and then scroll up to CustomRGB and change the gamma to 1.0, give it a custom name and click OK and then scroll up to Save RGB... where you'll save it as a regular .icm profile.

               

              Now make a duplicate of the original TIFF posted here and convert one to regular CIERGB and the other to the newly made custom 1.0 gamma profile. Then in curves apply to both images these settings:

               

              Output box 135 and Input box 175 to create a curve that darkens the image similar to your posted samples. Reverse the entries to create a lightening curve. The entries have to be the same for both images.

               

              You should get two different colored versions as I did that are even more different than the ones in your sample.

              • 4. Re: Why curves in ACR acts so different in comparison on other programms?
                amam123 Level 1

                Yes of course, Tim, I got different images, which have same color primaries but different gamma. However tendency still the same: while darkening - skin tones shifts to red, and while lightening - shifts to yellow. Major differences only in density. In LR we gets _reverse_ tendency. This is very uncomfortable while make adjustment in different programms (LR+PS) on same photo. My brains have to be switched while photo transfer from LR to PS Therefore I do not touch curves, contrast and exposure brightness correction in LR...

                I wonder that only adobe programmers implemented this algorithm while the rest world (even in photoshop) another.

                Where I can make this question to adobe if not here?

                • 5. Re: Why curves in ACR acts so different in comparison on other programms?
                  Tim Lookingbill Level 1

                  Not sure where you can get an answer to this. Maybe Thomas Knoll, the author of Photoshop and ACR, can take a stab at it.

                   

                  There are so many image processing workflows developed by a wide range of users, I imagine Adobe engineers expected most to process Raw files (metadata edited based images) in ACR/LR and tiffs (gamma encoded pixel based images) in Photoshop.

                   

                  I've wondered about the same thing myself just out of curiosity but don't find an issue with it in my workflow. I work in Raw exclusively now and find the smooth and much refined movements with very little noise in the previews I get editing Raw in a linear wide gamut color space far exceed what I get editing a TIFF version in Photoshop that I do all edits in ACR as close to finished as possible where I never need to apply such a drastic curve shown in those samples.

                   

                   

                  I don't understand why you'ld want to switch back and forth between ACR/LR and Photoshop editing TIFFs when you can just do everything in Photoshop, but then I'm not familiar with your workflow.

                  • 6. Re: Why curves in ACR acts so different in comparison on other programms?
                    amam123 Level 1

                    I shot often in available incandescent light plus sometime low mixed light and want to keep slight warm tint to photos. In case if I a want to achieve "low-key" look I have to make these drastic adjustments. Therefore I get faces more yellow and dirty in shadows instead of pleasant reddish hue, which I easily get from any other programm instead of LR.

                    I want to do all adjustments in LR and do not use PS completely because of huge amount of photos I take in session. Finishing photos by editing via tiff and using actions or filters will slows down my work and will make doubling disk space which is not desirable to me.

                    For comparison: Capture One satisfies me completely in terms of quality and LR in speed. But now I prefer speed instead of quality. I would like to take speed from LR and quality from C1 .

                    Since PS have different layer modes (luminosity, saturation, etc...), different gamma spaces almost do not influence on result picture.

                    • 7. Re: Why curves in ACR acts so different in comparison on other programms?
                      Tim Lookingbill Level 1

                      Sounds like you're using and comparing other image editors as a profiling device for your camera that just happens to make your edits fit your taste in color. That's not a recipe for consistency.

                       

                      I'm also assuming you're shooting theses images as you've indicated under mixed non-uniform spectra light sources expecting to get a warm rendering by clicking for R=G=B which as you've seen doesn't work. There's no way you can get consistent results easily in getting that specific warm hue shooting under changing spectra light sources. You're going to have to do some adjustments on every image. You might try quick adjustments to the orange and/or yellow sliders in the Hue panel in LR's HSL panel. That's what I have to do to get that reddish hue to skin.

                       

                      Why not profile the camera shooting Raw under these lights using an X-rite color chart and the DNG Profile Editor or some other camera profiling package? Integrated Color has a wizbang camera profiler but it cost around $500. Check out the sample image comparisons between converters at this site:

                       

                      http://www.integrated-color.com/cecamera/

                       

                      They seem to lean toward a reddish rendering with better definition out of the box which isn't accurate but is pleasing.

                       

                      Also you might try Lightroom 3 Beta which has a new preview rendering algorithm which might include the color effect you see in Photoshop.

                       

                      Here's the forum for it:

                       

                      http://forums.adobe.com/community/labs/lightroom3

                      • 8. Re: Why curves in ACR acts so different in comparison on other programms?
                        amam123 Level 1

                        >Sounds like you're using and comparing other image editors as a profiling device for your camera that just happens to make your edits fit your taste in color. That's not a recipe for consistency.

                        Simply I often used curves in PS and when LR has appeared I expected same behavior of curves. I have mistaken.

                        I see that many PS actions like TRA or Kubbota`s uses this property of PS curves (skin tones darkening shifts to red, lightnening shifts to yellow) to make photos great/pop looking. But I do not want to use PS. Again it requires making aside tiff.

                         

                        >You're going to have to do some adjustments on every image.

                        Yes. But most light still incandescent: concerts, private rooms, cafes.

                         

                        >You might try quick adjustments to the orange and/or yellow sliders in the Hue panel in LR's HSL panel. That's what I have to do to get that reddish hue to skin.

                        Yes, I do this, but when I touch curves on opposite side I have to adjust hue sliders again. This is not very quick. But for this case I have made many presets of HSL sliders for different curves and simply apply them

                         

                        >Why not profile the camera shooting Raw under these lights using an X-rite color chart and the DNG Profile Editor or some other camera profiling package?

                        I will repeat again (do not take offence): Curves still act the same way in spite of right colors achieved by this calibration.

                         

                        >They seem to lean toward a reddish rendering with better definition out of the box which isn't accurate but is pleasing.

                        I will read about this technology. Thank you.

                        In case I have to get correct/right colors now I use self made icc profiles by X-Rite Colormunki device and freeware profiling software (argyll) and custom made color checker with at least 108 patches instead of standard 24. But I have to follow these conditions: 1. Same light. 2. Same settings in developing software (that means I cannot touch any curve, contrast, hue and other sliders therefore why to use LR in general? ). 3. Developing program must operate with icc profile (sadly LR can't do this).

                         

                        Some addition information about right and pleasant colours: Now I have Canon`s SLR camera, but very love nikon`s colours (camera profiles). By dng profile editor I have applied for my cr2 files nikon's camera profiles and have got truly nikon`s colors without buying nikon camera. Colours exactly the same as in Capture NX in default settings until I start to adjust contrast, curves in LR...

                         

                        >Also you might try Lightroom 3 Beta which has a new preview rendering algorithm which might include the color effect you see in Photoshop.

                        I have tried LR3 for a long time ago but see it still have same algorithm of curves and contrast. May be developers add in final release some color manage preferences for choosing working colour space or so.

                        • 9. Re: Why curves in ACR acts so different in comparison on other programms?
                          amam123 Level 1

                          Ok, lets see comparison on test studio shot to avoid my "bad" light conditions.

                          Make attention on tone separation on hair. LR completely made them in one tone, but in PS I see nice separated riches tones.

                          How I can to reddish hairs in central picture to achieve result as in right picture in LR by HSL sliders? I tried, but I has failed.

                          In other programms I gets almost same result as in PS.

                          • 10. Re: Why curves in ACR acts so different in comparison on other programms?
                            Bill_Janes Level 2

                            Flyshow wrote:

                             

                            I shot often in available incandescent light plus sometime low mixed light and want to keep slight warm tint to photos. In case if I a want to achieve "low-key" look I have to make these drastic adjustments. Therefore I get faces more yellow and dirty in shadows instead of pleasant reddish hue, which I easily get from any other programm instead of LR.

                            If you are attempting to white balance with curves in a gamma 2.2 space, you will have problems. The proper tools here are the temperature and tint sliders. ACR curves do operate on the linear raw data, but the results are what you would get for the gamma of the working space. Why don't you post a problematic raw file and the problem rendering? Others might suggest how you could get the deisred results.

                            • 11. Re: Why curves in ACR acts so different in comparison on other programms?
                              Jeff Schewe Level 5

                              Flyshow wrote:

                               

                              "Does anybody knows, why Adobe made adjustment of curves in lightroom so different in comparison of curves Photoshop?

                               

                              Very simply...when Thomas Knoll was developing Camera Raw, he made intentional changes to the curves algorithms in Camera Raw to eliminate what he though was a problem in Photoshop's curves process. Photoshop will alter not only the luminance but also the saturation AND the hue when adjusting curves. In Camera Raw the curves function modify primarily the luminance data and slight adjustments to saturation but not hue adjustments. And you prolly want to be using parametric curves not the overall simplified brightness and contrast in ACR/LR.

                               

                              So, the two processes are fundamentally different on purpose...

                               

                              Your problem is you seem to have a singular use case where the ACR functions doesn't "help you" like the Photoshop curves function. Such is life...if you have a specific need that ONLY Photoshop can do, do it in Photoshop unless you can learn to do it in ACR/LR.

                               

                              Learning how to expertly use the tools you are using makes a lot more sense than wishing the tools worked differently and be frustrated...

                               

                              The simply fact is, neither Camera Raw nor Lightroom will be changing any time soon...so, either YOU will have to learn how to use ACR/LR or do stuff in Photoshop.

                              • 12. Re: Why curves in ACR acts so different in comparison on other programms?
                                amam123 Level 1

                                Thank you, Jeff. I completely satisfied with this answer.

                                • 13. Re: Why curves in ACR acts so different in comparison on other programms?
                                  Tim Lookingbill Level 1

                                  Well said, Jeff.

                                   

                                  This discussion got me to test this color effect in ACR on a Raw shot of my own hand under a quite neutral GE Sunshine 5000K fluorescent tube. I could retain the yellow in the orangy brown tan portion of my hand by pulling my hand away from the light to darken it as if I applied a curve similar to what's been demonstrated here and it retained this yellow cast just as it did darkening the shot in ACR. There was no physical increase or hue shift to red/orange viewing my actual hand under these lights.

                                   

                                  Doing this on the extracted jpeg version from the same Raw shot, I noticed this yellow affect amplified compared to the Raw. The thing is my jpegs have a duotone style white balance baked into the color table of its rendering where the mids have a tinge of green compared to the highlight's warmer hue.

                                   

                                  Including a WhiBal card in the shot proved this because when I clicked on the gray patch of the card to get R=G=B, I kept getting readings of the white patch of around 240,230, 210. I could not click in this white area to get R=G=B in ACR.

                                   

                                  In the Raw version of the shot I could click anywhere, white or gray patch, and get R=G=B for both patches.

                                  • 14. Re: Why curves in ACR acts so different in comparison on other programms?
                                    amam123 Level 1

                                    Tim, in summary of this discussion I understand then I wanted nice/pleasant colours in my own taste instead of mathematically right behaviour of ACR curves.

                                    Of course I will try and learn how achive pleasant results in ACR. But to be true I still see hue shifts when adjusting contrast.

                                    I tried to compare ACR with Caprure NX. CNX on the same tiff file in LCH mode truly do not shifts any hue at all. Plus in CNX I have ability to choose method (RGB or lightness). This is more comfortable to photographer. ACR have limits in this case.

                                    • 15. Re: Why curves in ACR acts so different in comparison on other programms?
                                      Romain_Th

                                      It seems like you are talking about the hue twists generated by the profiles. These are deliberate features of the various profiles (except the one you could have done with a colorchecker in the DNG profile maker): they modify hue and saturation depending of the luminance.

                                      Example of the benefit you get: a bright sky will get slightly darker and more saturated but the same blue at a lower luminance (ex: a swimming pool) won't get changed.

                                      Check out that guy's blog, he is explaining the thing and also provide a tool to "untwist" the profiles if you want. You can also build your own with a colorchecker that won't have twists.

                                       

                                      http://chromasoft.blogspot.com/2009/02/adobe-hue-twist.html

                                       

                                      I hope this helps.

                                       

                                      Romain

                                      • 16. Re: Why curves in ACR acts so different in comparison on other programms?
                                        amam123 Level 1

                                        I know what does dcptool. But In test I used tiff file to avoid influences of camera profiles.