15 Replies Latest reply on Nov 11, 2007 11:20 AM by

    Increased harshness in ACR 4.2

      I'm working in PS CS3 and using ACR 4.2 on a Mac. I noticed that the files got a somewhat harsh graininess that I could possibly describe as "rice grains" and that details seemed smeared in an artificial way that I don't recall from previous versions of ACR. It's particularily unpleasant in faces and skin tones.
      So I opened the same .CR2 image from my Canon Eos 20D in CS (ACR 2.4), CS2 (ACR 3.7) and CS3 (ACR 4.2) with basicaly the same settings, no sharpening and a minimum of noise reduction.
      The result is that this graininess seems to increase and get harsher for every new version of ACR.
      The only side-effect that I can consider as positive is that ACR 4.2 automaticaly smears/removes "hot pixels" from exposures of 30 seconds or more.
      Could I please get any comments on the suposed benefits of this?

      /*Peter Rosén
        • 1. Re: Increased harshness in ACR 4.2
          michael shaffer Level 1
          Peter ... terms like graininess seems to increase and harsher are subjective terms -- whereas it is so true about a picture being worth a thousand words. It would help a lot if you could send us to a webpage or embed 1:1 crops that show what you are trying to describe.

          I thought I'd be able to provide a link that would let you know how this is done, but I cannot find it ... anyone?
          • 2. Re: Increased harshness in ACR 4.2
            Level 1
            Hello Michael,
            I totaly agree. I have prepared a file (150K) to upload if anyone can tell me how to do it. I am also willing to send it as an attachment to an e-mail if there is no other way to do it.

            /*Peter
            • 3. Re: Increased harshness in ACR 4.2
              Dirk Williams Level 1
              Peter, We really need to have the raw file to look and test with. Or do you have a side by side for us to look at?

              You can send it to me if you'd like: monsterpup at lycos dot com
              • 4. Re: Increased harshness in ACR 4.2
                Level 1
                Hello Dirk,

                I have sent you an e-mail with a side by side example of what I am suggesting.

                Thanks
                /*Peter
                • 5. Re: Increased harshness in ACR 4.2
                  Dirk Williams Level 1
                  Hi Peter, When I try to open the files It's saying the files are corrupt or the file format is not recognized... except for CS2_CS3Comp-1 and CaptOneComp.

                  Looking at the CS2_CS3Comp-1 Jpeg, the two side by side images at the top look almost identical except for what looks like some added sharpening that might of been added to the one on the left (ACR 3.7) image. The one on the right (ACR 4.2) looks a little smoother. Where I see the difference is on the right shoulder of the shirt. The one on the right looks to have a little more contrast.

                  Are you sure that in ACR 3.7 the sharpening is turned off or/and at zero? Also what ISO are you shooting at? It almost looks like (Looking at the bottom two) I see a little noise from shooting at a higher ISO.
                  • 6. Re: Increased harshness in ACR 4.2
                    Level 1
                    Hi, I only sent 2 files, so there is no problem with corruption.
                    This particular file was shot at 400 ISO but I really don't think it's relevant.
                    I prepared this example just for the purpose of showing my point, but it is true for any file at any ISO setting, at least from Canon Eos 20D or 5D. I am certain that sharpening was off in all 3 figures. My point is the strange "ricegrain" shaped noise that gets stronger in ACR 4.2 than in ACR 3.7 when sharpening the image, particularily visible and undesirable in skin tones. Other tests I have made show that it was even less in ACR 2.4.
                    My impression is that there is some increased smearing on pixel level implied in ACR 4.2 resulting in bigger blobs which get quite visible and unpleasant when a necessary sharpening is applied to the image.
                    I know that in such an issue it is easy to get stuck in technicalities so it would be nice to hear from someone implied or who has first hand knowledge of the programming of ACR, if there is any point or fact in what I'm stating.

                    Thanks

                    /*Peter Rosén
                    • 7. Re: Increased harshness in ACR 4.2
                      Mathias Vejerslev Level 2
                      Peter,

                      If you read through this list, you'll see this 'complaint' repeated over and over. I believe the 'graininess' in 4.2 stems from the updated demosaicing algorithm, possibly coupled with the new sharpening / denoising masks.

                      Why mess with the demosaicing? Because its one of the premier image quality factors. For instance, the increased graininess might translate to better luminosity detail than before. Which do you prefer?

                      Try comparing Canons DPP to Camera RAW, and you'll find that DPP generates blanket 'graininess' all over the photo, including 'blank' background areas, whereas ACR has it much better under control.
                      • 8. Re: Increased harshness in ACR 4.2
                        Mathias,
                        This works to a point. That point being getting to the higher ISOs and how the demosaicing handles the noise and detail in images as it presents it to the remainder of the tool (ACR or LR) for work. At lower ISOs I have very little issue. At higher ISOs, the results of the demosaicing create the issues many have described. I think this has been evidenced by the fact the some have changed the ISO of RAW images going in (from 1600 to 100) for example and have stated a better result. For obvious reasons the more the noise and detail combined in the RAW file, the more difficult to strike a balance one would suppose.

                        Jay S.
                        • 9. Re: Increased harshness in ACR 4.2
                          Level 1
                          Mathias,
                          I do definitively prefer lesser graininess and to work up luminosity detail in PS, as it used to be. I want to be able to start with an image that looks and feels as natural and with as little artifacts as possible. Artifacts will always be added upon editing the image (sharpening, adding saturation etc) but then at least I can control it. It's very dificult to go the other way and remove artifacts while keeping a natural photographic look.
                          Canons mid-range SLR cameras already feature an anti-alias filter that is too strong to my liking so there is no room for increased smearing in the raw converter. I think that Canons top line SLRs as well as Nikon and possibly other brands have a less extreme approach to in-camera anti-aliasing and noise reduction.
                          I don't know anything about DPP as I have never tried it. I used to be quite satisfied with ACR before this last upgrade. So I can only hope that the ACR team will go back to a more conservative approach in the future or find myself another raw converter that suits my needs better.

                          /*Peter
                          • 10. Re: Increased harshness in ACR 4.2
                            Mathias Vejerslev Level 2
                            Maybe what we really need is a 'detail' slider á la Raw Shooter Essentials, so we can draw our own compromise?
                            • 11. Re: Increased harshness in ACR 4.2
                              Mathias,
                              Do you mean like the Detail slider in Lightroom? In RSE/RSP, it was called Detail extraction. In the Processing Bias aspects of RSP, there was something called Detail Extraction vs. Noise Reduction which w as nice, but I'm not sure what the equivalent would be in ACR or LR.

                              Jay S.
                              • 12. Re: Increased harshness in ACR 4.2
                                Level 1
                                Peter, you might want to read the following topics (there are more):

                                http://www.adobeforums.com/webx/.3bca1cbc/0
                                http://www.adobeforums.com/webx/.3bc44a00/0

                                and particularly
                                http://www.adobeforums.com/webx/.3bc45755/0
                                with lost archive at http://www.adobeforums.com/webx/.3c0347de/249

                                If this is your problem, a change in 4.3 is promised. Whether for good we shall see.

                                ivos
                                • 13. Re: Increased harshness in ACR 4.2
                                  Level 1
                                  Thanks Ivo for your posting,
                                  I have now read some of the topics you are linking to and indeed it is to the point what I am refering to, just more eloquently described by others and by you. I noticed the problem on files shot at 100 - 400 Iso and also the watercolor and "Panasonic" like effect. This is absolutely not what I want to end up with when shooting with my Eos cameras.
                                  While waiting for the next release of ACR I will have a look at DPP and a second look at CaptureOne 4,0 beta that seemed more to my liking at a first quick evaluation.

                                  /*Peter
                                  • 14. Re: Increased harshness in ACR 4.2
                                    Mathias Vejerslev Level 2
                                    Jay,

                                    I mean like Detail Extraction. It is not like the detail slider in Lightroom / ACR, but rather some kind of demosaicing parameter - acuteness with artifacts, or less artifacts and less acuteness.
                                    • 15. Re: Increased harshness in ACR 4.2
                                      Mathias,
                                      I am trying to draw any co-relation, as I said between those RSP sliders and anything in LR, and I just don't think it is there. From what I had set up, the Detail Extraction vs. Noise Reduction on RSP (in the Processing Bias, had the more significant impact on High ISO Noise.

                                      Jay