1 2 Previous Next 42 Replies Latest reply on Aug 22, 2010 10:18 AM by Noel Carboni

    Easy capture sharpening in ACR 6

    RASouthworth Level 3

      After reading most of the books available on ACR and sharpening, including Jeff's latest, and after having dinked with this stuff for many years, I've come to a startling conclusion, at least to me.  Setting Amount = 150, Radius=1. Detail = 0, and Masking = 0 seems to work well for CAPTURE sharpening on all of my images.  And I'm including low and high iso, "high frequency" and portrait, etc.

       

      I capitalized capture because to me these settings work well on the initial image prep, narrowing the light/dark transitions without adding any discernable halo.  Doing as is generally recommended, bringing up Detail to add a slight halo visible at 400% but not at 100%, with corresponding Amount and Radius adjustments, can certainly produce a slightly sharper appearing image on the monitor, but IMO one has now at least partially crossed over into output sharpening for the monitor.

       

      Another advantage of the 150-1-0-0 approach is the lack of any artifacts as one continues to process the image, particularly upsampling for printing.  For noisy images I do end up adjusting Masking as I interatively apply NR, but again in general the settings work well.  And of course setting them up as ACR defaults simplifies processing.  I'm curious as to whether others have arrived at a similar conclusion, or not.  I haven't been doing this as a default very long, and wonder if I'll come across pitfalls later.

       

      Richard Southworth

        • 1. Re: Easy capture sharpening in ACR 6
          ACRFREAK Level 1

          Funny enough but I am a landscape photographer/fine art print maker (lightjet silver halide) and I find almost the opposite to work the best for my images.  I generally have the amount at 30-45, the Radius at 0.5, the Detail at 100 and the masking slight (anywhere from 0 to maybe 30).  I too read all the books very carefully.

          • 2. Re: Easy capture sharpening in ACR 6
            RASouthworth Level 3

            My understanding is setting the Detail to 100 causes the ACR sharpening to be very close to the unsharp mask in Photoshop, i.e. the radius then sets the width of the "halo" and the amount sets the strength.  I'm not saying this is wrong, it's very close to what I do prior to printing an image, although then the threshold control also comes into play.  However it would not be my choice for initial capture sharpening, wherein the intent is to compensate for the inherent slight softening caused by demosaicing the sampled image.

             

            As usual there are many opinions as to the best way to prepare the image, and I'm not trying to say I've found the best approach.  I just found my "discovery" interesting and wondered if others had arrived at a similar conclusion.  The best way to observe the process is to bring an image into ACR, zoom it to 400%, and set the sharpening to 150-1-0-0.  Now move the Amount slider back and forth while examing a light/dark edge, such as a eye highlight.  It seems to "clean" up edges without creating the halos normally found in sharpened images examined at high zoom levels.  Also back off the zoom to 100% and repeat the experiment, still does a noticeable crisping up of the image with no crunchiness.

             

            Richard Southworth

            • 3. Re: Easy capture sharpening in ACR 6
              Bill_Janes Level 2

              RASouthworth wrote:

               

              After reading most of the books available on ACR and sharpening, including Jeff's latest, and after having dinked with this stuff for many years, I've come to a startling conclusion, at least to me.  Setting Amount = 150, Radius=1. Detail = 0, and Masking = 0 seems to work well for CAPTURE sharpening on all of my images.  And I'm including low and high iso, "high frequency" and portrait, etc.

               

              Jeff's latest book has some misinformation, which is surprising since he claims to have been involved in the development of the ACR sharpening process. In a LuLa thread, Eric Chan confirmed that the ACR uses a blend of unsharp masking and deconvolution sharpening, with the blend determined by the Detail slider of ACR. A detail of 100 uses pure deconvolution.

               

              http://www.luminous-landscape.com/forum/index.php?topic=45038.msg378027#msg378027

               

              Jeff states that using a detail of 100 produces results very similar to what would be obtained by an unsharp mask, but my experience indicates otherwise. Sharpening can be evaluated with Imatest using a slanted wedge to measure the line spread function and MTF. I photographed a target photographed with the Nikon D3 and rendered with ACR 6.1. I used a radius of 0.7, since I am interested in landscape work and that is a good radius for such work. Here is the result using ACR with a radius of 0.7, Amount of 125 and Detail of 0. Contrary to what Jeff said, there is slight overshoot.

               

              003_125_0p7_0_YL26_cpp.png

               

              And here is ACR with a radius of 0.7, Amount 125, and Detail of 100. There is marked overshoot.

               

              003_125_0p7_100_YL26_cpp.png

               

              And here is an image unsharpened in ACR and sharpened with the Unsharp mask in Photoshop using a radius of 0.7 and amount of 125. The results are similar to that obtained with ACR Radius 0.7, Amount 125, and Detail 0:

               

              003_UnSh0_7_125_YL26_cpp.png

              To avoid marked ringing artifact with Detail of 100, one must use a relatively small. Eric Chan recommended Detail of 100 for landscapes. A Detail of 50 or less is best in this situation:

               

              003_50_0p7_100_YL26_cpp.png

              Of course, with an a real landscape image, one would judge the result onscreen at 100%, but Imatest lets you see what is actually going on.

              • 4. Re: Easy capture sharpening in ACR 6
                Jeff Schewe Level 5

                Bill_Janes wrote:

                 

                Jeff's latest book has some misinformation, which is surprising since he claims to have been involved in the development of the ACR sharpening process. In a LuLa thread, Eric Chan confirmed that the ACR uses a blend of unsharp masking and deconvolution sharpening, with the blend determined by the Detail slider of ACR. A detail of 100 uses pure deconvolution.

                 

                 

                That's what Eric said, true...but that's not what Mark Hamburg said when we were developing the Camera Raw sharpening back in ACR 4.1. I know Eric has done a lot to enhance what Mark and I worked on...but the post he made on LuLa was the first I've heard that he has made such changes to the high frequency component of the Detail slider. I'm sure to get the full scoop because I'll be spending a week with Eric at an upcoming workshop. I'll report back what I discover...

                 

                As to the OP, yes, there are a lot of potential settings in the Detail panel that might be appropriate for a given image. However, to blindly use an Amount of 150 (the amount goes to 150 because of me) with a Radius of 1 with Detail and Masking to 0 is, I think leaving potential image quality on the table. You also need to factor in Luminance Noise Reduction (almost any image can benefit from a slight amount of reduction, even native IOS captures with a good exposure). I personally suspect you've fallen in love with a "look" and are accepting perhaps less than you could pull out of your images if you don't bother to use 2 of the arguably 5 sliders used for detail enhancement in ACR. But of course, that's just me. If you're happy, I'm tickled pink.

                • 5. Re: Easy capture sharpening in ACR 6
                  Bill_Janes Level 2

                  Jeff Schewe wrote:

                   

                  That's what Eric said, true...but that's not what Mark Hamburg said when we were developing the Camera Raw sharpening back in ACR 4.1. I know Eric has done a lot to enhance what Mark and I worked on...but the post he made on LuLa was the first I've heard that he has made such changes to the high frequency component of the Detail slider. I'm sure to get the full scoop because I'll be spending a week with Eric at an upcoming workshop. I'll report back what I discover.

                  Jeff,

                   

                  I suspect that the sharpening has changed since ACR 4.1. I no longer have that version on my computer and it may not even support my current camera. I did repeat my Imatest evaluatiion using Radius 0.75, Amount 75, and Detail 100 using the 2003 process and the 2010 process. The results are quite different.

                   

                  Process 2003:

                   

                  003_75_0p7_100_2003_YL26_cpp.png

                   

                  Process 2010:

                   

                  003_75_0p7_100_YL26_cpp.png

                   

                  Regarding the OP's method of Amount 150, radius 1.0 and Detail 0, I don't think that workflow would be optimum since he is using the unsharp mask sharpening method which merely gives the impression of sharpness whereas using Detail would employ deconvoluiton, which actually improves detail (pun intended). Please do report back what you learn from Eric.

                  • 6. Re: Easy capture sharpening in ACR 6
                    RASouthworth Level 3

                    Bill,

                     

                    I'm an EE, so I do understand your terminology and curves.  However, I'm still somewhat confused as to your use of unsharp and deconvollution.  I though unsharp implied the existence of a halo, however slight, which does not occur with Detail = 0.  If I'm reading your curves correctly I believe they confirm my conclusion, i.e. Detail=0 results in little overshoot/ringing, and therefore no halo.  And again, I'm not proposing 150-1-0-0 as the best setting for all images, just as a good start for batching large amounts of images thru without adding artifacts.

                     

                    Jeff,

                     

                    I don't believe "blindly" is a fair assessment, I'm doing this after a fair amount of examination, and I understand that improvement can be wrung out of individual images by further adjustment.  And I do have Luminance set to 20 as a default, having found as you state that almost all images can stand some help with NR.

                     

                    So again I ask not if this is "best" for all images but if used stricly as a capture sharpening technique will it cause problems after further image editing.

                     

                    Richard Southworth

                    • 7. Re: Easy capture sharpening in ACR 6
                      RASouthworth Level 3

                      As usual a picture is worth a thousand words, here is a screen shot illustrating the effects of ACR 6.2 (2010 process) sharpening using Amount=150, Radius=1, Detail=0, Masking=0.  Again this seems at least adequate for capture sharpening, particularly since it is mostly free of any artifacts that might cause problems in later editing.

                       

                      The 100% and 400% on the left are without any sharpening, the right hand pair are with 150-1-0-0.

                       

                      http://rgbaustin.com/SharpTest.jpg

                       

                      Richard Southworth

                       

                      Added by edit - I guess there is no way to view at 100% within the forum, please download the image and view within PS to examine at full resolution.

                      • 8. Re: Easy capture sharpening in ACR 6
                        ACRFREAK Level 1

                        To my eye that is a nice example.  I see where you are coming from.  Extract detail but attempt to do so without halo damage,  That has definitely got my attention.  Time to do more tests!  I'll be standing by as the issue get pushed further.  Thank you for bringing it up.

                        • 9. Re: Easy capture sharpening in ACR 6
                          ACRFREAK Level 1

                          OK, well I just tried this on a landscape image that is basically almost all high/mid frequency:

                           

                          Amount 150

                          Radius 0.5

                          Detail 0

                          Masking 0 (although I would mask appropriately with any image)

                           

                          and then compared it to

                           

                          Amount 35

                          Radius 0.5

                          Detail 100

                          Masking 0

                           

                          Critical viewing was at 100, 200, 300 and 400% in ACR.

                           

                          I am really surprised at this!  Although the sharpening is extremely close to the same amount in appearance (at 100%) the first settings with detail at 0, to my eye has a lot cleaner edges (less halowing) and definitely less artifacting in smoother areas (even w no masking).  Basically at the edges of the image there seems to be less feathering of pixels transitioning them (may interpolate up better this way?).

                           

                          I will do some more testing, to see how the file then handles Sharpening for Output (with the new datail at 0 settings).  I am thinking at this point that your detail at 0 idea might be sound. Less artifacting and cleaner edges seems to mean that the file will take more aggressive sharpening after up sizing it, and then the grain simulation I use to disguise the artifacting may be able to be a little less aggressive.

                           

                          Question, do you still approach the masking (ACR) the same way?

                           

                          Lastly, I tried using the first settings of sharpening on an image interpolated to max size in ACR (about 20MP) and then tried the same capture sharpening settings on an image at native size (about 12MP) then interpolated up to 20MP using PS Smoother (remembering that ACR applies the sharpening after the uprez). I have found the ACR uprez to have more detail at ultra zoom levels.  But if ACR applies the capture sharpening after the ACR uprez (if my reading is right, Jeff S said that) and the other is capture sharpened at native pixel size and then uprezzed with Bicubic Smoother, aren't I just comparing apples with oranges?  Also, when taking a image from 12.9 to 19.5MP is the ACR interpolating using its algorithm that is like Smoother?

                          • 10. Re: Easy capture sharpening in ACR 6
                            RASouthworth Level 3

                            I have my defaut Luminance set to 20, which cleans up most images up to ISO 800 or so, from a D300.  If I start working with more NR I will bring up masking approriately.  I haven't run long enough with this scheme to have a real good handle on using with high ISO images.

                             

                            Richard Southworth

                            • 11. Re: Easy capture sharpening in ACR 6
                              ACRFREAK Level 1

                              Richard, question:  why radius at 1.0? 

                               

                              Do you see any reason I would get better results at that, then say 0.5 on my shots which are usually 100 iso, landscape, high to mid frequency (I often entirely mask out the real soft stuff like sky/long exposure water with a layer below with NO Capture Sharpening).

                               

                              So the question is why 1.0 instead of 0.5?

                              • 12. Re: Easy capture sharpening in ACR 6
                                RASouthworth Level 3

                                All I can say is try different values and watch at 100% and 400%.  I found little or no effect going to lower radius, and if much more it seem to lose the sharpening.  Again, this is all at Detail = 0.

                                 

                                Richard Southworth

                                • 13. Re: Easy capture sharpening in ACR 6
                                  Bill_Janes Level 2

                                  RASouthworth wrote:

                                   

                                  Bill,

                                   

                                  I'm an EE, so I do understand your terminology and curves.  However, I'm still somewhat confused as to your use of unsharp and deconvollution.  I though unsharp implied the existence of a halo, however slight, which does not occur with Detail = 0.  If I'm reading your curves correctly I believe they confirm my conclusion, i.e. Detail=0 results in little overshoot/ringing, and therefore no halo.  And again, I'm not proposing 150-1-0-0 as the best setting for all images, just as a good start for batching large amounts of images thru without adding artifacts.


                                  Richard,

                                   

                                  By unsharp, I am referring to the usual unsharp mask. Deconvolution is a complicated subject and you might look at the article by Roger Clark where he compares the Richardson-Lucy deconvolution algorithm to the unsharp mask.

                                   

                                  http://www.clarkvision.com/articles/image-restoration1/

                                   

                                  You are correct that the regular unsharp mask has a halo and this is suppressed in ACR with a detail of zero. However, the ACR sharpening with a radius of 100, Amount 150 and Detail does introduce alaising and other artfacts. Another deconvolution method readily available on the PC as a Photoshop plugin is FocusMagic. I sharpened my image with ACR 6.1 (process 2010) with your settings and with FocusMagic using a blur width of 1 and amount of 175. Raw conversion for the FM was done in ACR with the same settings except for no sharpening. The Imatest analyses are shown.

                                   

                                  ACR 6.1 Sharpening R = 1, Amount = 150, Detail = 0

                                  003_150_1_0_YL26_cpp.png

                                   

                                  Focus Magic, Blur width = 1, Amount = 175

                                   

                                   

                                  003_FM_1_175_YL26_cpp.png

                                   

                                  The ACR sharpening shows a very high MTF at the Nyquist frequency (0.5 cycles/pixel), which often indicates alaising. To prevent alaising at frequencies above Nyquist, most dSLRs have a low pass filter, while these are omitted for most medium format digital backs. Alaising on images can be mistaken for sharpness, and the effects of alaising on images are hotly debated.

                                   

                                  Alaising with your settings is readily evident when one examines the converted images at 100%. Viewers should double click on the iamge to view it at 100%. The FM is on the left and the ACR with your settings on the right. Note increased Moire at the higher frequencies on the upper right and increased alsising on the parallel bars on the bottom with your settings. The FM does have overshoot, but this in itself is not unfavorable if no halo is evident. Overshoot does increase edge contrast. Personal preferences will vary and evaluation of sharpening is highly subjective, but I prefer the FM.

                                   

                                  Comp2.png

                                  • 14. Re: Easy capture sharpening in ACR 6
                                    RASouthworth Level 3

                                    Bill,

                                     

                                    Thanks for the information, very interesting.  And you've answered my question as to when my simple-minded approach can cause a problem

                                     

                                    I do believe it sharpens better for most images than the ACR defaults, so my plan is to leave 150-1-0-0 as the default until I run into an image that shows up the weakness.  I did search thru images for some with an obvious fabric patterns, fences, etc. trying to invoke an artifact, but was not able to find such.  Just have to stay alert.

                                     

                                    Richard Southworth

                                    • 15. Re: Easy capture sharpening in ACR 6
                                      Bill_Janes Level 2

                                      RASouthworth wrote:

                                       

                                      Bill,

                                       

                                      Thanks for the information, very interesting.  And you've answered my question as to when my simple-minded approach can cause a problem

                                       

                                      I do believe it sharpens better for most images than the ACR defaults, so my plan is to leave 150-1-0-0 as the default until I run into an image that shows up the weakness.  I did search thru images for some with an obvious fabric patterns, fences, etc. trying to invoke an artifact, but was not able to find such.  Just have to stay alert.

                                      I agree that that the ACR defaults are too weak, but they show little artifacts or alaising. Your settings are reasonable, but I have to agree with Jeff Schewe that you may be leaving some image quality on the table by a one size fits all approach. The resolution of the camera and strength of the low pass filter affect the settings for capture shapening, as does the image content: e.g. landscape vs portraits.

                                       

                                      I think you could get better results by decreasing the amount and adding a bit of deconvolution (i.e. moving detail to the right). Here are examples of my preferred settings for the D3 and landscape.These images must be viewed at 100% to avoid on screen alaising.

                                       

                                      ACR Amount 75, Radius 0.75, Detail 25

                                      75_0p7_25_Comp.png

                                       


                                      Image Magic, Blur width = 1, Amount 175

                                      FM_1_175.PNG

                                       

                                      Raw Therapee with Richardson Lucy deconvoluiton at default settings

                                      RL_comp.png

                                       

                                      I can't see any alaising on the RL image, in agreement with the low MTF at Nyquist, but noise is increased (ISO 400 was used). The FM does show a bit of alaising, and ACR with the chosen settings shows a bit more alaising. For most purposes, I favor using ACR to preserve parametric editing. I would be interested in what others are using and what Jeff learns in his meeting with Eric Chan. Eric suggested 100% deconvolution with ACR.

                                      • 16. Re: Easy capture sharpening in ACR 6
                                        RASouthworth Level 3

                                        Bill,

                                         

                                        Again thanks for the information and analysis, all very fascinating.

                                         

                                        I set up two ACR 6.2 presets so I could do quick A-B testing on images from my D300/16-85VR combination, one with your settings of 75-.8-25-0 (how were you able to achieve .75, my copy kept rounding off?) and one with 150-1-0-0.  On a well lit outdoors portrait shot I was able to see a slight improvement at 400% with your settings, at 100% no discernable difference.  I brought up a direct sun house shot, and with your settings at 400% I saw a slight halo or echo by a high contrast white/black vertical edge, although just barely visible, that did not occur with my settings. Other than the slight artifact I could not see any difference in overall image "sharpness" at 400%.  At 100% the very slight artifact was not visible with your settings, and of course the two settings produced visually identical results.

                                         

                                        As both you and Jeff have pointed out one can wring out more detail with custom settings for each image, but I remain convinced for a simple "one size fits all" approach, e.g. running thru several hundred vacation shots, the 150-1-0-0 combo works well.  And of course I now have another preset (with your name embedded within) as an alternative.

                                         

                                        Richard Southworth

                                         

                                        Added by edit - went back and examined some more "high frequency" building shots; the 75-.8-25-0 setting accentuated some noise (as well as being slightly sharper in general) in the lower midtones, say 60-60-60 or so, whereas the 150-1-0-0 combination did not.  In both cases I had Luminance=20.  The effect was minor and only visible at 400%.

                                        • 17. Re: Easy capture sharpening in ACR 6
                                          ACRFREAK Level 1

                                          I appreciate this thread! 

                                           

                                          I have been doing experimenting/testing with this (on my landscape, very big enlargements) and am pretty much convinced I am getting more detail with less artifacting, and doing less damage to the file at the raw stage, using a similar recipe. I think there seems to be something to be said about not haloing, or keeping the halos substantially constrained in relationship to causing"damage."  In other words, the more haloing done in raw (it seems to me) the less aggressive I can later sharpen the image for output when pushing enlargement sizes.

                                           

                                          In a simple sense, what I see here is a substantially tightened up image in raw (the goal of undoing the softness that naturally occurs when you convert a steady stream of photons into discrete squares) but without the typical, sort of unnatural (if you will) halo "damage." My theory is that since the halos arn't there, they are not later getting exacerbated.

                                           

                                          Mine is:

                                           

                                          Amount 150

                                          Radius 0.5  (I am not seeing a benefit of 1.0 on my work yet, and am trying to mainly pull out the fine detail)

                                          Detail 0 (I am entertaining the idea of testing moving this slightly - like 0-5 maybe - when a more aggressive sharpening is needed OR working in the 0.5 to 1.0 range in Radius instead)

                                          Masking (image dependant)

                                           

                                          I should note that on smooth surfaces like clouds or long exposure water I just bring in another image with NO Capture Sharpening, to mask in for those areas in PS.

                                          • 18. Re: Easy capture sharpening in ACR 6
                                            Bill_Janes Level 2

                                            ACRFREAK wrote:

                                             

                                             

                                            I have been doing experimenting/testing with this (on my landscape, very big enlargements) and am pretty much convinced I am getting more detail with less artifacting, and doing less damage to the file at the raw stage, using a similar recipe. I think there seems to be something to be said about not haloing, or keeping the halos substantially constrained in relationship to causing"damage."  In other words, the more haloing done in raw (it seems to me) the less aggressive I can later sharpen the image for output when pushing enlargement sizes.

                                             


                                             

                                            Amount 150

                                            Radius 0.5  (I am not seeing a benefit of 1.0 on my work yet, and am trying to mainly pull out the fine detail

                                            Detail 0 (I am entertaining the idea of testing moving this slightly - like 0-5 maybe - when a more aggressive sharpening is needed OR working in the 0.5 to 1.0 range in Radius instead)

                                            Mea culpa. In my post #13 in this thread my Imatest results for ACR sharpening with a radius = 1, Amount = 150, and Detail = 0 were for process 2003 and not process 2010. With these settings in process 2003, there is no overshoot (which could cause halos), whereas with Process 2010 there is overshoot (similar to my previous posts at other settings). Results for a radius of 0.5 are similar, but less marked. With process 2010, the halo is not pinned completely at Detail = 0, just as the behavior at Detail = 100 is not similar to results obtained with the unsharp mask (as Jeff Schewe states in his ACR books). Previous explanations of ACR sharpening such as found in Jeff's ACR books need to be updated.

                                             

                                            ACR 6.1 Process 2010, Amount = 150, Radius = 1, Detail = 0

                                             

                                            003_150_1_0_YL26_cpp.png

                                             


                                            For a DSLR with a blur filter, better results are likely from a deconvolution method. See the LuLa thread on this topic:

                                             

                                            http://www.luminous-landscape.com/forum/index.php?topic=45038.0.

                                             

                                            Eric Chan knows something about ACR and he often uses Detail = 100 for landscapes, which invokes deconvolution sharpening. To prevent severe ringing, one has to use a relatively small amount. Anyway, I have belabored this point enough and would appreciate input from others.

                                            • 19. Re: Easy capture sharpening in ACR 6
                                              MadManChan2000 Adobe Employee

                                              While I do use a very high Detail (50 to 100), low Radius, and low-to-moderate Amount, for my personal landscape work, I do not recommend it as a general-purpose starting point.

                                               

                                              This is because the appropriate setting of Detail and Radius (which define the "character" of the sharpening) depend on the image content and shooting conditions. For my non-landscape work, where a smoother result with less "bite" is desired, I tend to use Detail more in the first half of its range (e.g., 10 to 25 for portraits, a bit more than that for street photography, etc.)

                                              • 20. Re: Easy capture sharpening in ACR 6
                                                RASouthworth Level 3

                                                Eric,

                                                 

                                                So just what is occurring when sharpening with 150-1-0-x?  I've tried this with many images from both my D300 and S90, with consistent results, i.e. "pleasing" edges with no apparent artifacts.  And I realize that for individual images more detail can be brought out with custom settings, but again I haven't found anything better for a "one size fits all" approach.

                                                 

                                                Richard Southworth

                                                • 21. Re: Easy capture sharpening in ACR 6
                                                  Bill_Janes Level 2

                                                  MadManChan2000 wrote:

                                                   

                                                  While I do use a very high Detail (50 to 100), low Radius, and low-to-moderate Amount, for my personal landscape work, I do not recommend it as a general-purpose starting point.

                                                   

                                                  This is because the appropriate setting of Detail and Radius (which define the "character" of the sharpening) depend on the image content and shooting conditions. For my non-landscape work, where a smoother result with less "bite" is desired, I tend to use Detail more in the first half of its range (e.g., 10 to 25 for portraits, a bit more than that for street photography, etc.)

                                                  Eric,

                                                   

                                                  Thanks for your input. From my experience and testing, there seems to have been a considerable change in ACR sharpening in recent versions, where higher detail invokes deconvoluiton. Jeff Schewe has reported in his ACR books that Detail of 100 produces results very similar to what would be obtained by the unsharp mask and that a detail of zero "will almost completely pin the sharpening edge halo". In my testing with Imatest and ACR 6.1 Process 2010, a detail of 100 with large amounts produces marked halos and these are reduced with lesser Amount values. However there is still some overshoot at Amount = 0, in contrast to Process 2003. If even Jeff Schewe is not fully aware of these recent changes, the rest of us are even less informed.

                                                   

                                                  I understand that you and Jeff will be attending a conference soon, and the results of your consulations would be of great interest. Some general understanding of exactly what the sliders do helps to use them intelligently. Of course, no one expects you to divulge proprietary algorithms. I think that the new shaprening in ACR is an improvement and applaud your work on them, but I think better documentation would be helpful.

                                                  • 22. Re: Easy capture sharpening in ACR 6
                                                    ACRFREAK Level 1

                                                    Updated!

                                                     

                                                    As I state in my first comment on this thread, I too have been using an approach much like Mad Man on all my landscape (which is what I do for a living) fine art enlargements (24 - 70+ inches) mainly DSLR.  I have been using ACR 6.1 and the 2010 rendering.  I too came up with the idea that I wanted to pull out more of the nitty gritty detail (which led me to try those previous settings like Mad Man).

                                                     

                                                    I may be wrong (see updated below because I am) but my test the other day was 35 (and 34) - 0.5 - 100 - 0  VS  150 - 0.5 - 0 - 0.

                                                     

                                                    At 100% the detail looked (subjectively) just about identical in overall sharpness but with less artifacts at the new 150 - 0.5 - 0 - 0.  At 400% the immediate thing I noticed was, where did all the halos go?  This was a mid to high frequency Aspen Trees shot with a Canon 1DS Mark II (older image) at f/11 with the 24-105.

                                                     

                                                    I then did an uprez on both and what seems obvious to me was that although the two were relatively the same in visual sharpness at 100% before the uprez, after the Bicubic Smoother uprez the one capture sharpened at 150 - 0.5 - 0 - 0  could take a more aggressive amount of USM (in LAB L) than the other. The end result of the test was that the 150 - 0.5 - 0 - 0  looked sharper to my eye with less weirdness (artifacting).

                                                     

                                                     

                                                    Update :

                                                     

                                                    I just did another totally diferent image test of a slot canyon wall (extremely high frequency only - sandy walls) and it failed miserably!  On that one the best result was 40 - 0.5 - 100 - 0.  The 150 - 0.5 to 1.0 - 0 - 0 looked terrible in comparison.  It was soft looking and all the really fine sand detail pretty much wiped out!

                                                     

                                                    Conclusion, it totally depends on the images characteristics and frequency levels!

                                                     

                                                    Would be nice if one day we had a sharpening algorithm that detected frequencies accurately and them applied just the right type of sharpening to just the right frequencies!

                                                    • 23. Re: Easy capture sharpening in ACR 6
                                                      ACRFREAK Level 1

                                                      Update :

                                                       

                                                      I just did another totally diferent image test of a slot canyon wall (extremely high frequency only - sandy walls) and it failed miserably!  On that one the best result was 40 - 0.5 - 100 - 0.  The 150 - 0.5 to 1.0 - 0 - 0 looked terrible in comparison.  It was soft looking and all the really fine sand detail pretty much wiped out!

                                                       

                                                      Conclusion, it totally depends on the images characteristics and frequency levels!

                                                       

                                                      Would be nice if one day we had a sharpening algorithm that detected frequencies accurately and them applied just the right type of sharpening to just the right frequencies!

                                                      • 24. Re: Easy capture sharpening in ACR 6
                                                        Level 4

                                                        ACRFREAK wrote:

                                                         

                                                        …Conclusion, it totally depends on the images characteristics and frequency levels!

                                                         

                                                         

                                                        That's what the gurus here have been saying all along, and that has always been my experience.

                                                         

                                                         

                                                        ACRFREAK wrote:

                                                         

                                                        …  The 150 - 0.5 to 1.0 - 0 - 0 looked terrible in comparison.  It

                                                         

                                                        I had been scratching my head in utter disbelief and incomprehension  since "the 150 - 0.5 to 1.0 - 0 - 0" was first mentioned here. 

                                                         

                                                         

                                                        ACRFREAK wrote:

                                                         

                                                        …image test of a slot canyon wall (extremely high frequency only - sandy walls)…  On that one the best result was 40 - 0.5 - 100 - 0…

                                                         

                                                        That sounds totally reasonable and more in step with reality as I've encountered it..

                                                         

                                                         

                                                         

                                                        Wo Tai Lao Le

                                                        我太老了

                                                        • 25. Re: Easy capture sharpening in ACR 6
                                                          RASouthworth Level 3

                                                          Tai Lao,

                                                           

                                                          other than scratching your head in "utter disbelief and incomprehension" have you actually tried the 150-1-0-x settings on images?  I never claimed it was the answer for all, but I still have not found an image in my set (D300/16-85VR or S90) that did not benefit.  Rather than throwing stones perhaps a little experimentation and reporting back might help in the thread progress.

                                                           

                                                          ACRFreak,

                                                           

                                                          I'm interested in seeing the results on your canyon shot, and what was the camera/lens combination?  And I'm digging back into my image set trying to find one similar to yours.

                                                           

                                                          Richard Southworth

                                                          • 26. Re: Easy capture sharpening in ACR 6
                                                            Level 4

                                                            RASouthworth wrote:

                                                             

                                                            …other than scratching your head in "utter disbelief and incomprehension" have you actually tried the 150-1-0-x settings on images?

                                                             

                                                            Yes, of course.  I tried it after reading your post.  That's when the head scratching started!  

                                                            • 27. Re: Easy capture sharpening in ACR 6
                                                              RASouthworth Level 3

                                                              Ok, and what were the results?  Did it improve some images and deteriorate others?  Were the results at 100% and 400% signficantly worse/better than other settings?  Any images to show?

                                                               

                                                              Again, most of my examinations were with the D300/16-85VR combination, in general starting with properly focused images, all thru ACR 6.2 running 2010 process.

                                                               

                                                              I'm also not sure as to what constitutes proper "capture" sharpening.  It appears that some believe one should go for the "sharpest" results at 100% at native resolution with no visible halo even if that means visible artifacts at 400%.  I have believed, perhaps erroneously, that capture sharpening should result in little or no artifacts at any resolution, so that at the end of the editing process one can apply output sharpening with minimal problems.  And of course output sharpening can range from adjustments to a down-rez'd image for monitor viewing or an up-rez'd image for printing.  It almost appears to me that the more common "capture" sharpening definition has become optimize the image at 100% pixel viewing, which in general is not going to be the final output resolution.

                                                               

                                                              Richard Southworth

                                                              • 28. Re: Easy capture sharpening in ACR 6
                                                                Level 4

                                                                RASouthworth wrote:

                                                                 

                                                                Ok, and what were the results?  Did it improve some images and deteriorate others?

                                                                 

                                                                Tried just a few images to satisfy my curiosity, portraits, architectural shots, product shots and landscapes.  The results were simply "outlandish" to say the least, so I moved on.

                                                                 

                                                                In general, I'm pretty satisfied with PhotoKit Sharpener in Photoshop.

                                                                 

                                                                 

                                                                Wo Tai Lao Le

                                                                我太老了

                                                                • 29. Re: Easy capture sharpening in ACR 6
                                                                  RASouthworth Level 3

                                                                  Tai Lao,

                                                                   

                                                                  I wonder if we're using the same settings.  I posted an example of my experience earlier in this thread, and the results were certainly not "outlandish" as you can see for yourself.  Again, Amount=150, Radius=1, Detail=0, and Masking=0, ACR 6.2, 2010 process.  And I'm not the only poster to have achieved reasonable results with this combination.

                                                                   

                                                                  I would think you would be curious as to why your results would be so much different than mine.  Basic sanity testing says there must be a disconnect somewhere.  I guess if someone is satisfied with their present workflow then anything else could be considered nonsense.

                                                                   

                                                                  Richard Southworth

                                                                  • 30. Re: Easy capture sharpening in ACR 6
                                                                    Level 4

                                                                    RASouthworth,

                                                                     

                                                                    This is by no means worth pursuing.  At least to me.

                                                                     

                                                                    For one thing, I'd have to go and borrow my colleague's machine again to do more tests.  I'm on CS4 on my own machine.

                                                                     

                                                                    My intention was not to argue or to debate, simply to state that your proposed settings seemed weird to me, so I went out and tried them on a machine that runs ACR 6.2, and the results left me scratching my head.  That's all.

                                                                     

                                                                    Someone else in this thread seems to have had a similar experience.

                                                                     

                                                                     

                                                                    Wo Tai Lao Le

                                                                    我太老了

                                                                    • 31. Re: Easy capture sharpening in ACR 6
                                                                      ACRFREAK Level 1

                                                                      I think the main point here is that for a one size fits all, quick auto - default approach your 150 - 1.0 - 0 - ?  may be an option for some.  Indeed, on image I tested those settings did produce a superior result than any other approach (althoug I have Radius at 0.5).   But it produced a much worse result now on two images I have tested.  So when sheer quality is the aim, the results are image dependant and thus I have now abandoned the idea of using it as a default setting.

                                                                      • 32. Re: Easy capture sharpening in ACR 6
                                                                        Noel Carboni Level 7

                                                                        Not long ago I went through the process, with the modern ACR 6, to see if I could discern a good set of defaults for converting my images, taken with my camera, and knowing how I am likely to process them once converted.

                                                                         

                                                                        There is SO much involved with this outside of ACR, and SO many controls even in ACR, that it's impossible to find "one way" or even "one best way" to work.  I defy anyone to even begin to argue otherwise.  Add to that the dimension of tweaking the controls for particular images and changing developed image processing and it gets even MORE difficult to nail down.

                                                                         

                                                                        In short:  To each his own.  One man's trash is another man's treasure.  That sort of thing.

                                                                         

                                                                        In my own case, in order to make the most detailed possible images from my Canon 40D images I chose to pursue a "convert to higher than native resolution" approach involving some sharpening during conversion.  It was a time consuming process, but I tried methodically changing settings, processing the images in a way I typically do, then critically examining the results for a "better/worse" overall reading.

                                                                         

                                                                        I ended up going in an interesting and somewhat unexpected direction:  Where in the past I always preferred conversions with no sharpening and minimal noise reduction, now I ended up dialing in Amount: 15 sharpening at Radius: 0.5 with Luminance: 40 noise reduction and Luminance Detail:100.  I also chose Color noise reduction: 4, Color Detail: 100, and I now think I have a decent understanding of what I should tweak in individual exposures to make things like higher ISO images more pleasing.  Started with my defaults, I have already had several people use the term "Wow!" when viewing my images - I couldn't ask for more.

                                                                         

                                                                        I'm not sure there's any quick way around this mountain of adjustability.  You have to get to know your particular hardware and software (and computer capacity) and make your own judgments about how you want your images to look, and do some experimentation of your own.

                                                                         

                                                                        -Noel

                                                                        • 33. Re: Easy capture sharpening in ACR 6
                                                                          Level 4

                                                                          Noel Carboni wrote:

                                                                           

                                                                          …There is SO much involved with this outside of ACR, and SO many controls even in ACR, that it's impossible to find "one way" or even "one best way" to work.  I defy anyone to even begin to argue otherwise…

                                                                           

                                                                          I agree 100%.

                                                                           

                                                                          Noel Carboni wrote:

                                                                           

                                                                          …I ended up going in an interesting and somewhat unexpected direction:  Where in the past I always preferred conversions with no sharpening and minimal noise reduction, now I ended up dialing in Amount: 15 sharpening at Radius: 0.5 with Luminance: 40 noise reduction and Luminance Detail:100.  I also chose Color noise reduction: 4, Color Detail: 100, and I now think I have a decent understanding of what I should tweak in individual exposures to make things like higher ISO images more pleasing.  Started with my defaults, I have already had several people use the term "Wow!" when viewing my images - I couldn't ask for more…

                                                                           

                                                                          That sounds perfectly reasonable and would not leave me scratching my head at all.

                                                                           

                                                                           

                                                                          Wo Tai Lao Le

                                                                          我太老了

                                                                          • 34. Re: Easy capture sharpening in ACR 6
                                                                            Yammer Level 4

                                                                            I've been through several "phases" of settings in ACR, where I've changed the defaults, only to change them back again several months later.

                                                                             

                                                                            Sharpening is one group which I'm still playing with. About a year or so ago, I was using 50-60 as default sharpening amount, and doing minimal sharpening in PP. This gave me problems with longitudinal CA halos, as well as unsatisfactory results in certain images. I ended up dropping back to Adobe defaults.

                                                                             

                                                                            With the new PV2010, I've been experimenting with higher values again. But I've noticed problems with subsequent resampling, so I'm being very careful to only sharpen images destined for print, leaving JPEGs at default and sharpening only after resampling.

                                                                            • 35. Re: Easy capture sharpening in ACR 6
                                                                              RASouthworth Level 3

                                                                              In my experience the "key" to using large values of Amount is to use very small values of Detail, in many cases 0.  While not the best settings for all images it works fairly well (at least for me) to achieve a reasonable amount of Capture sharpening with little or no artifacts and therefore minimal problems with later editing/sharpening.

                                                                               

                                                                              And if you try it properly, at least on ACR 6.2 Process 2010, it won't leave you scratching your head at all.

                                                                               

                                                                              Richard Southworth

                                                                              • 36. Re: Easy capture sharpening in ACR 6
                                                                                Bill_Janes Level 2

                                                                                RASouthworth wrote:

                                                                                Again, most of my examinations were with the D300/16-85VR combination, in general starting with properly focused images, all thru ACR 6.2 running 2010 process.

                                                                                 

                                                                                I'm also not sure as to what constitutes proper "capture" sharpening.  It appears that some believe one should go for the "sharpest" results at 100% at native resolution with no visible halo even if that means visible artifacts at 400%.  I have believed, perhaps erroneously, that capture sharpening should result in little or no artifacts at any resolution, so that at the end of the editing process one can apply output sharpening with minimal problems.  And of course output sharpening can range from adjustments to a down-rez'd image for monitor viewing or an up-rez'd image for printing.  It almost appears to me that the more common "capture" sharpening definition has become optimize the image at 100% pixel viewing, which in general is not going to be the final output resolution.

                                                                                 

                                                                                Richard,

                                                                                 

                                                                                I don't think your criterion that proper capture sharpening should produce no artifacts at 400% is reasonable. The unsharp mask is old technology, but works by producing sharpening halos around edges. As Bruce Fraser explained in his Real World sharpening book where he shows examples at 300%, proper sharpening produces halos which are sufficient to cause the impression of better detail but are not visible at output resolution. If you want to avoid sharpening overshoot, you should use Process 2003, which appears to completely dampen them at Amount = 0.

                                                                                 

                                                                                A digital image at 400% looks terrible, and no one in his right mind would print at anywhere that resolution. 400% is fine to see what is going on in the sharpening process, but many artifacts at that resoluiton would likely not show up at any reasonable printing resolution.

                                                                                • 37. Re: Easy capture sharpening in ACR 6
                                                                                  RASouthworth Level 3

                                                                                  Bill,

                                                                                   

                                                                                  I well understand that 400% viewing is in general not useful as to assessing final print quality, I was only pointing to it as a method to determine what degree artifacts are produced.  And I also understand that some halo from capture sharpening may not cause any difficulties for final print output, although I'm not convinced that avoiding them "penalizes" the image in its final usage.

                                                                                   

                                                                                  But it appears to me that the term "capture sharpening" means different things to different people.  I read Bruce Fraser's book, even went thru his entire process on a couple of images, capture/effect/output sharpening with the masking etc. to see the results (they were good but it was a painful process).  I guess I wonder if my simple minded approach to capture shapening puts my images at a disadvantage in terms of the final output quality, given that at the end of editing and re-sizing I properly apply output sharpening.

                                                                                   

                                                                                  Richard Southworth

                                                                                   

                                                                                  Added by edit - I do apprediate yours and others input to this thread, has been very educational - there are obviously many more tools available today for image processing, at all stages.  It was not too long ago that one used the raw converter to do only that, and capture sharpening consisted of the Unsharp Mask set at radius between .5 and 1.0, threshold at zero, and amount to taste, with more manipulation required for noiser images.  Now we have a sophisticated (and evolving) set of functions within the raw converter, in addition to more options within Photoshop.  Makes for interesting times (or is that a curse?).

                                                                                  • 38. Re: Easy capture sharpening in ACR 6
                                                                                    ACRFREAK Level 1
                                                                                    function(){return A.apply(null,[this].concat($A(arguments)))}

                                                                                    .

                                                                                                                                             

                                                                                     

                                                                                    And if you try it properly, at least on ACR 6.2 Process 2010, it won't leave you scratching your head at all.

                                                                                     

                                                                                    Richard Southworth

                                                                                     

                                                                                    Again, totally depends on the image.

                                                                                     

                                                                                    On some images I have tested this on it leaves me on my knees with my head in the toilet.  On some images (like the aforementioned Aspen Landscape shot) it is the best thing since cotton candy.  Since it varies so wildly, as a fine art landscape photographer (not production) I can not accept it as any type of default setting.  But to be honest, I will check out the approach on my images now.  For that I appreciate this post.

                                                                                    • 39. Re: Easy capture sharpening in ACR 6
                                                                                      Noel Carboni Level 7

                                                                                      For what it's worth, I have found that viewing an image at 50% on the screen can give a decent feel for how it will look in print at high ppi.

                                                                                       

                                                                                      -Noel

                                                                                      1 2 Previous Next