11 Replies Latest reply on Aug 16, 2010 11:46 AM by ACRFREAK

    Posterization in ACR?

    ACRFREAK Level 1

      I shoot mainly Canon DSLR landscape and sell large high end Silver Halide gallery prints..  Lately I think I have been finding (rarely) some indication of slight posterization in some very smooth transition areas in ACR 6.1 raw (case in point a sunset sky).  I'm basically talking about where the pixels don't have, but should have a very smooth tonal transition to them).  The images I am talking about were exposed optimally (histogram to the right) and no super radical settings applied. Sometimes it is not immediate in ACR but starts happening very quickly as I go into PS (16 bit) and start making some adjustments.

       

      Has anyone else seen this?  If so, please feel free to share.

       

      It caught me by surprise because for some reason I thought ACR CR2 raw files (and even 16 bit) could not do this.

       

      I also have found (another totally different issue) that strong use of reversed "Clairity" (the glow/smoothing effect) seems to be able to exacerbate posterization as well.  Often when it is used somewhat aggressively in a localized area and then brought into PS (16 bit) there can be some (what looks to me like) posterization that no amount of healing brush cloning or blurring... resolves (again even in 16 bit).  There does seem to be a correlation between how strong the ACR adjustments and the amount of posterization is.

       

      I thought maybe it was some sharpening issue (artifacts created by) but I have tried turning the capture sharpening off (and there is no other sharpening applied) and all sharpening off and I can still see it.

       

      When I tried "Neat" (in PS noise reduction plug in) to see if that would eliminate it, it made it much worse.  I have found that a touch of grain simulation can disguise/camouflage it OK.

       

      Maybe I'm off here, but I am just wondering if anyone else has experienced this or a similar thing?

        • 1. Re: Posterization in ACR?
          Jeff Schewe Level 5

          So, are you seeing the banding in Camera Raw or Photoshop? What EXACTLY did you do to the images in Photoshop? Do you see the exact same banding in print that your see on the display?

           

          Fact is, I've never seen banding in Camera Raw. I have seen it crop up in Photoshop if you are doing adjustment layers with modified gradients–meaning lay out a gradation and then apply a curve to the layer mask.

           

          The odds are, what you are seeing is display banding caused by the profile of your display. Do you "calibrate" your display or merely profile it at the native white point and gamma?

          1 person found this helpful
          • 2. Re: Posterization in ACR?
            ACRFREAK Level 1

            Just like I expect from you Jeff, well thought out answer with good questions, thank you.

             

            I think I am seeing slight banding in ACR 6.1, maybe.  On closer examination just now, I think it might be pixels where chrominance noise was resolved (no luminance noise reduction used at all) and so my statement about perfect exposure may not be so perfect (100 iso on the Canon 1DS Mark II and moderate noise in ACR says to me I had to brighten the area up too much).

             

            As far as the same banding in print, no.  But my last test print had a touch of grain sim to try to disguise it.  What I hear you saying here is to do a test print without the "grain" and see if there is anything visible...

             

            "Fact is, I've never seen banding in Camera Raw."  That is a reassuring statement.  I mainly wanted to throw this out to see if posterization is something people have seen in ACR, or should I look for another culprit.

             

            But, yes when I use a lot of the "Anti Clarity" (as I call it) in an area (in ACR 6.1) and the area is an extremely slow transition area and then I bring the image into PS (CS5) in 16 bit it seems the area is very prone to posterization quickly.

             

            "The odds are, what you are seeing is display banding caused by the profile of your display. Do you "calibrate" your display or merely profile it at the native white point and gamma?"

             

            I never thought of that possibility.  I just run the Spyder 2.2 and then use that profile.  I do BTW need to get up to speed on color management/calibration due to getting extremely anal about printing quality these days.  All suggestions are welcomed!

             

            Thanks you.

            • 3. Re: Posterization in ACR?
              Jeff Schewe Level 5

              ACRFREAK wrote:

               

              I never thought of that possibility.  I just run the Spyder 2.2 and then use that profile.  I do BTW need to get up to speed on color management/calibration due to getting extremely anal about printing quality these days.  All suggestions are welcomed!

               

              If you are seeing banding in Photoshop (and not in Camera Raw or in prints) the logical place to look would be your display profile. Not that if you force a white point AND a gamma adjustment when calibrating and profiling your display, you will no longer have a full 8-bits/channel at the display. I've often seen banding on displays that were there in real life for exactly that reason. Which is why I don't "calibrate" my displays, I only profile them in their native white point and gamma. Of course, the displays I use are NEC SprectraView displays with 10 bit adjustments at the display (not in the pipeline, but in the display) so I don't need to push a white point or gamma adjustment using the display calibration software. When I set the display to D65 and gamma 2.2, the display itself does an internal adjustment. Then all I need to do is profile the state of those settings.

              • 4. Re: Posterization in ACR?
                Noel Carboni Level 7

                Something to consider:  To the best of my knowledge in Photoshop proper, a selection mask - however feathered and in whatever image color depth you work - is ALWAYS 8 bit.  If such a selection is used as part of your workflow (e.g., a highly feathered selection of a smooth object or whatever), posterization can creep in.  This does not apply to layer masks, just selections.

                 

                I say this because I develop plugins, and I know a selection mask is ALWAYS received by a plugin in 8 bit depth, regardless of the other factors.

                 

                -Noel

                • 5. Re: Posterization in ACR?
                  Jeff Schewe Level 5

                  Noel,

                   

                  Correct...selections are limited 8-bit precision. Channels and layer masks in a 16-bit/channel file ARE 16 bits (well, 15 bit plus 1 level). That's why it's a real good practice to use layer masks instead of trying to work within an active selection if you can. This is one thing that has bothered me since Photoshop added more 16-bit functionality back in version 6 or 7 (or whenever it was). I wish they would update selections to be 16-bit in 16-bit documents...

                  • 6. Re: Posterization in ACR?
                    ACRFREAK Level 1

                    HOLY Smokes that is a great tip!!!  I am a masking/selection master freak!

                    • 7. Re: Posterization in ACR?
                      JimGoshorn Level 2

                      So if you have a mask and make it a selection, it drops down from 16-bit to 8-bit precision?

                       

                      Jim

                      • 8. Re: Posterization in ACR?
                        Jeff Schewe Level 5

                        JimGoshorn wrote:

                         

                        So if you have a mask and make it a selection, it drops down from 16-bit to 8-bit precision?

                         

                        Yep...

                         

                        If you use the mask as a layer mask then you get the benefit of full 16 bit precision.

                        1 person found this helpful
                        • 9. Re: Posterization in ACR?
                          ACRFREAK Level 1

                          Although I am not going to sweat it, why did no one tell me this years ago!!!!

                           

                          So, hypothetically/conceivably a person could be degrading locally (in areas) to the point of posterization because of 8 bit areas being introduced into the 16 bit space.  Time to mask...

                          • 10. Re: Posterization in ACR?
                            Jeff Schewe Level 5

                            In the grand scheme of things, the 8 bit selection isn't necessarily a huge deal. Most of the time, feathering a selection to get a soft edge is done with a relatively low radius of a couple of pixels. In that case, the limitations of 8 bit selection is largely irrelevant. Where an 8 bit selection can potentially cause problems is where you convert a 16 bit mask or channel into an 8 bit selection. A large radius feather or a gradation based selection could become an issue.

                            • 11. Re: Posterization in ACR?
                              ACRFREAK Level 1

                              So, maybe that is why I see banding sometimes in the selection preview of Refine Edge (I use the "on black" mask preview option) when Smooth is 100%, feathering fairly strong and Shift edge contracted.  I always wondered if that actually translated into banding of the live selection.  Maybe it does.