19 Replies Latest reply on Sep 3, 2010 7:48 PM by bmb-sea

    Why are ACR tone & exposure tools contaminating my channels?

    bmb-sea Level 1

      I have a digital scanning back camera that outputs DNG files.

       

      These files require an adjustment on a per channel basis (shifting etc). But I cannot do that until I open the DNG file and then save it out again to tiff.

       

      Any changes made to the file (curve for example) in ACR, shift data from one channel to the other. Thus preventing me from doing the proper additional corrections to each of the channels afterwards because the channels are now contaminated with data from other channels.

       

      The curve tool in Photoshop does not do this, nor does its exposure tool or its levels tool. If you make a change in the curves etc, it effects the channels in parallel without shifting color from one channel to another. This is how I need it to perform in ACR.

       

      Why does ACR do this? How can I have ACR leave my channels separate when I make these types of adjustments?

        • 1. Re: Why are ACR tone & exposure tools contaminating my channels?
          Noel Carboni Level 7

          As far as I know, there is no adjustment in ACR that allows you to change tone curves for individual color channels.  The exposure slider and tone curves in ACR by definition affect all the channels.

           

          Are you saying the adjustments you're using are unexpectedly affecting hue and/or saturation, so that if you had an image with data only in one channel (say, an image shot entirely in red light), that after making adjustments you'd see data in the other channels that wasn't there before?  In short, how are you seeing/testing this?

           

          -Noel

          • 2. Re: Why are ACR tone & exposure tools contaminating my channels?
            bmb-sea Level 1

            As far as I know, there is no adjustment in ACR that allows you to change tone curves for individual color channels.  The exposure slider and tone curves in ACR by definition affect all the channels.

             

            Yes I am aware of that.

             

            Are you saying the adjustments you're using are unexpectedly affecting hue and/or saturation, so that if you had an image with data only in one channel (say, an image shot entirely in red light), that after making adjustments you'd see data in the other channels that wasn't there before?

             

            Correct, this is exactly what I am saying.

             

            In short, how are you seeing/testing this?

             

            You can see it yourself. To make it more apparent what is happening, do the following:

             

            1. open any photo in photoshop and shift each of the channels (R,G & B) different amounts horizontally.

            2. save two copies each as a tiff file

            3. open one tiff file in ACR (you can do this through Bridge) and apply a curve for example

            4. then click "open" in ACR to get the photo back into photoshop

            5. shift each of the channels back again so they line up with each other.

            6. Notice the "cross talk" between channels.

             

            Now open the other tiff file you made with the offset channels in photoshop and apply a curve to it. Then shift the channels back so they line up again.

             

            Completely different result than what is produced by ACR. Each channel still has its own discreet data, uneffected by values from the other channels.

            • 3. Re: Why are ACR tone & exposure tools contaminating my channels?
              Noel Carboni Level 7
              function(){return A.apply(null,[this].concat($A(arguments)))}

              bmb-sea wrote:


              5. shift each of the channels back again so they line up with each other.

               

              I don't mean to be dense, but I don't understand exactly what you mean by "shift" a channel.  You mean move the channels one way or another by a horizontal or vertical offset?

               

              -Noel

              • 4. Re: Why are ACR tone & exposure tools contaminating my channels?
                bmb-sea Level 1

                You mean move the channels one way or another by a horizontal or vertical offset?

                 

                 

                Correct-

                 

                Menu->Filter->Other->Offset

                • 5. Re: Why are ACR tone & exposure tools contaminating my channels?
                  Noel Carboni Level 7

                  I guess what I'm not fully understanding here is why you would expect the manipulation of an image in its entirety (all channels) - given that the R, G, and B values of a given pixel are normally assumed to be related to one another - not to result in some combination of the data between the channels.  Perhaps Camera Raw is trying to work at a more "end user" level than what you expect.

                   

                  It just seems that a pleasing visual result is the key goal, perhaps at the expense of mathematical purity.  For example:  As a test I took a set of images at different exposure levels, and the Exposure slider did a marvelous job of making them look visually identical (ignoring changes in noise level).  I have no idea whether this involved coupling the channel data in the process, but I suspect it did.  Certainly the colors and saturations came out right.

                   

                  Then there's the camera profile, which defines how the (R, G, B) values from your camera are interpreted (forgive me for being a little shaky on how .dng files and camera profiles work together; I don't use .dng, so I'm not quite sure where the camera profile fits in your workflow).  The camera profile almost certainly combines channel data in its transforms.  Does a camera profile figure in your process, and if so, what profile are you using?

                   

                  Lastly, there's output color management to consider...  the (R, G, B) values for a "bright mauve" color might well have different proportions than those for a "dim mauve" color.  What document profile are you converting to?

                   

                  -Noel

                  • 6. Re: Why are ACR tone & exposure tools contaminating my channels?
                    bmb-sea Level 1

                    Hi Noel,

                     

                    There are technical reasons why I need ACR to function in the manner I need which I described in my first post. To further expound upon that would probably be beyond the scope/topic of this message board.

                     

                    I believe I have adequately described and shown how to see the issue which is preventing me from creating my final images in the manner that is required by the technology I use to capture them.

                    • 7. Re: Why are ACR tone & exposure tools contaminating my channels?
                      ssprengel Adobe Community Professional & MVP

                      Re-reading your first post, I am unsure if you are you wanting ACR to make channel-independent adjustments to the 1-color-per-pixel RAW data where two out of the three colors are necessarily computed from neighboring pixels of the other channels or whether you are using ACR to manipulate the 3-colors-per-pixel TIF data which is all that what Photoshop every has to deal with?

                       

                      In other words, can you at least specific whether you are manipulating DNG or TIF files with ACR when you see the problem you wish to avoid?

                      • 8. Re: Why are ACR tone & exposure tools contaminating my channels?
                        bmb-sea Level 1

                        Re-reading your first post, I am unsure if you are you wanting ACR to make channel-independent adjustments to the 1-color-per-pixel RAW data where two out of the three colors are necessarily computed from neighboring pixels of the other channels or whether you are using ACR to manipulate the 3-colors-per-pixel TIF data which is all that what Photoshop every has to deal with?

                         

                        I need to do the following type of correction to the file-

                         

                        See:

                         

                        http://www.all-in-one.ee/~dersch/pscan/pscan.html

                         

                        In other words, can you at least specific whether you are manipulating DNG or TIF files with ACR when you see the problem you wish to avoid?

                         

                        The camera outputs a DNG file. I cannot make edits to this file until I convert it to a TIFF. Converting this DNG file to TIFF in ACR contaminates the channels preventing me from doing the above described correction.

                         

                        This issue apparently applies to all files run through ACR, DNG or TIFF. Making adjustments in ACR effects the channels in ways that Photoshop (and most other programs) do not.

                        • 9. Re: Why are ACR tone & exposure tools contaminating my channels?
                          Noel Carboni Level 7

                          Could the Chromatic Aberration corrections in ACR possibly help with this, or is this unique because the image is captured in a scanning process?

                           

                          I suspect the CA corrections in ACR work so well on most raw images precisely because they're working on raw sensor data before many of the other machinations are performed.

                           

                          Perhaps you should request a special flavor of CA correction specifically for your kind of camera.

                           

                          -Noel

                          • 10. Re: Why are ACR tone & exposure tools contaminating my channels?
                            bmb-sea Level 1

                            Could the Chromatic Aberration corrections in ACR possibly help with this, or is this unique because the image is captured in a scanning process?

                             

                            No, ACRs built-in CA correction is a different type of channel shift from the primary adjustment that I need to make. Although I need to do CA correction to the files as well. However ACRs CA correction is applied radially in both directions.

                            Perhaps you should request a special flavor of CA correction specifically for your kind of camera.

                             

                            Yes, CA and falloff correction applied in only a single radial direction certainly would be nice to have.

                            • 11. Re: Why are ACR tone & exposure tools contaminating my channels?
                              ssprengel Adobe Community Professional & MVP

                              A scanning back DNG would have 3-colors-per-pixel, just like an RGB TIF, right, although there would still need to be a color-profile to describe the balance of sensitivities between the photosites just like a "regular" RAW needs, and unlike a TIF?  I am guessing that ACR is primarily designed for 1-color-per-photosite Bayer decoding which requires two of the three RGB colors to be guessed at each pixel, so there is some 1-CPP-specific optimizations that have not been quashed when a 3-CPP image is processed.

                               

                              Was part of your analysis, earlier, that processing a regular TIF through ACR has the same issue, so it's not only the RAW processing, but all processing that smears the channels?

                               

                              BTW, I would say that the technical details of your DNG corrections are completely appropriate for this forum because you want ACR to support corrections of a different type of camera "sensor" that has a unique set of issues to deal with.  It would be similar to discussion issues particular to the Sigma Foveon sensor where, besides 3-CPP, it has the added complication that the sensor is at a 45-degree angle.

                               

                              I think the request for Adobe for 3-CPP images would be that ACR perform processing w/o channel crosstalk, or at least have this be an option.  I also suspect that this is a question of priority of many items on the Adobe RAW-engine request/wish list and perhaps also cooperation from PhaseOne who is a competitor. 

                               

                              It would be interesting to hear from someone on the RAW engine team whether the smearing is part of the core decode-engine functionality, or part of a post-processing step that could more easily be tweaked for 3-CPP images.

                              • 12. Re: Why are ACR tone & exposure tools contaminating my channels?
                                bmb-sea Level 1

                                The DNG file from my scanback camera contains the exact same data that a 16bit linear TIFF would. The only difference is in the headers etc. that ACR uses to preprocess the file (curves). I suspect that other types of scanners will also produce these types of linear TIFF embedded DNG files.

                                • 13. Re: Why are ACR tone & exposure tools contaminating my channels?
                                  ssprengel Adobe Community Professional & MVP

                                  Sounds like ACR needs to try harder to keep things separate when the data is 3-CPP, already.

                                  • 14. Re: Why are ACR tone & exposure tools contaminating my channels?
                                    bmb-sea Level 1

                                    Perhaps there is a way to address from the opposite direction?

                                     

                                    Because apparently how ACR does its curves to the initial linear data in ways unlike Photoshop etc. may very well be why I prefer it as a starting point as opposed to opening the 16bit linear tiff in photoshop as a starting point (photoshop's curves are not meant for linear data).

                                     

                                    The image data in my 16bit tiff files and DNG files are the same. I can apply the corrections to the linear 16bit tiff I need to the different channels in other software I have specific for this purpose.

                                     

                                    So then the question can be-

                                     

                                    How can I get that "fixed" 16bit linear TIFF file back into ACR in a way that it can apply the same curves and other settings that it did when I bring in the DNG file? Is there way I can rewrap the modified TIFF with the same DNG wrapper my other file has? Remember, there is no de-mosaicing or other voodoo that ACR needs to deal with on these particular files. All the data is there and "ready to go" so to speak (same as with the DNG files from my camera).

                                    • 15. Re: Why are ACR tone & exposure tools contaminating my channels?
                                      ssprengel Adobe Community Professional & MVP

                                      The DNG Converter can make TIFs into DNGs, can't it, but I'm not sure it can make linearized DNGs, which have absolute color temperature, still, which I think is what you need.

                                       

                                      If you have a TIF corrected with the other software you speak of, could you post somewhere?  I think www.YouSendIt.com has a 200MB limit per file on the free version.

                                      • 16. Re: Why are ACR tone & exposure tools contaminating my channels?
                                        bmb-sea Level 1

                                        The DNG Converter can make TIFs into DNGs, can't it, but I'm not sure it can make linearized DNGs, which have absolute color temperature, still, which I think is what you need.

                                         

                                         

                                        Yes, that would pretty much fix my problem good however, I tried converting both linear tiffs to DNG again this way using the Adobe DNG converter. The uncorrected raw linear tiff and the corrected raw linear tiff both gave errors in DNG converter of "there was an error processing the file".

                                         

                                         

                                        If you have a TIF corrected with the other software you speak of, could you post somewhere?  I think www.YouSendIt.com has a 200MB limit per file on the free version.

                                         

                                         

                                        Yeah the files are huge. Ranging anywhere from 500MB to 1.5GB or so *each*. Maximum dimensions can be up to 9,000 x 60,000 pixels. There was a time a couple years ago when ACR wouldn't even open many of them (the larger ones).

                                         

                                        I will scale down one of the corrected linear tiff files and post it to my server if you would like to check it out.

                                        • 17. Re: Why are ACR tone & exposure tools contaminating my channels?
                                          bmb-sea Level 1

                                          Here is a link to a re-sized corrected linear TIFF-

                                           

                                          -removed-

                                           

                                          Btw, The above file is 1,000 x 2,000 (about 13 MB)

                                           

                                          Don't know if re-sizing it might have an effect on trying to get it back into DNG but there it is. I chose an image that had some vertical sensor blooming so the radial correction is more apparent (look at the top and bottom of the vertical blooming lines to see how the shifting of the channels was done).

                                           

                                          The way ACR applies its base curve gives me a better starting point than what I can achieve using Photoshop's curve tools (as they aren't really designed for dealing with linear data).

                                          • 18. Re: Why are ACR tone & exposure tools contaminating my channels?
                                            ssprengel Adobe Community Professional & MVP

                                            I'm not having any better luck than you with the TIF to DNG conversion.

                                            • 19. Re: Why are ACR tone & exposure tools contaminating my channels?
                                              bmb-sea Level 1

                                              Thanks for trying though! I hope to get this sorted out, there are some other people that are involved now.