1 2 3 Previous Next 83 Replies Latest reply on Aug 19, 2011 1:32 AM by Vit Novak

    Tweaking chromatic aberration correction

    Yammer Level 4

      Lens Correction in ACR 6.1 was a very welcome addition for me. I used to spend a lot of time fixing chromatic aberration, both manually and with Presets. I find that fixing lateral CA goes a long way towards improving image sharpness, so I'm a bit of a stickler for it. So, when 6.1 came along, I was keen so see what it could do for CA correction. I've been impressed, but, whilst CA is reduced, it's not removed, and I almost always have to fine-tune the correction manually.

       

      I have found that, even with pro lenses, I have to dial in significant (6–18) points of manual CA adjustment, and often very similar amounts. I have experimented with changing CA "correction amount" defaults, but it doesn't seem to help.

       

      So, three questions:

      1. is this expected behaviour, even for pro lenses?

      2. if so, will improved profiles ever be made available?

      3. is there any way I can "built in" offsets to lens profiles by default, to minimise having to use Presets or manual adjustment?

        • 1. Re: Tweaking chromatic aberration correction
          Hudechrome Level 2

          For default corrections, I suggest DXO. They are suppliers of software to the optics industry and as such, have used their technologies to build profiles for a great many lenses and bodies from the major manufacturers. They are automatic and so far as I have been able to discern, hit CA particularly well so that any manual tweaks, both in their app or ACR, degrades the auto correction.

           

          This isn't true for sharpness, but in this case, I assume that the practical sharpness on an individual frame not achieving optimum with DXO auto sharpness is due to more than inherent optical errors. That seems borne out by the fact that to achieve optimum sharpness in such frames, I have to use all tools I have available to get there. Sometimes ACR, sometimes DXO Manual, sometimes PS Smart sharpen. All three approach sharpening from divergent points of view, and if you include Pixel Genius, really divergent!

           

          IMHO, anyway.

          • 2. Re: Tweaking chromatic aberration correction
            Noel Carboni Level 7

            For what it's worth, YammerP, I often have to dial in some CA correction as well.

             

            In my case I use a lot of zoom lenses (some Canon L), and I've always suspected the CA amounts vary by the focal length some.  It's possible the CA correction profiles just don't take focal length of a zoom into account, and/or are simply slightly inaccurate.

             

            -Noel

            • 3. Re: Tweaking chromatic aberration correction
              Vit Novak Level 3

              Profiles are different. Some have several points for different focal lengths, and some even for various focal distances. Bigger the profile, more points

               

              So far, i tried profiles with kit lens for my 400D and my A650 compact. In first case, I used a profile for a new version of kit lens and corrected the profile manually, using a bunch of photos. It's a simple text file, with internal format described in a pdf file written by Eric and several other people, that I found somewhere ... In the second case, I used a profile for G10 and also corrected it manually. Works perfectly for me, although both lenses aver very inexpensive. Working on a SX110 profile at the moment - this is harder, as this lens is really pushed to the limit with big zoom range (huge amount of distortion and vignetting at some focal lengths), but it's almost finished

              1 person found this helpful
              • 4. Re: Tweaking chromatic aberration correction
                Yammer Level 4

                I just had a look at the Adobe profile for the Nikon 24-70mm f/2.8 pro lens, a "flagship" model. I usually have to add -7 to -15 in both Y/B and M/G manual correction on top of the automatic correction. The file appears to be XML, but it's HUGE (400kB) and seems pretty complex at first glance (too complex to attempt a manual adjustment. Maybe I need to look for alternative 3rd-party profiles.

                • 5. Re: Tweaking chromatic aberration correction
                  Vit Novak Level 3

                  Yes, in my case profiles were much smaller (only 3 or 4 different focal lengths), so I corrected the profiles manually, along with developping raws from last vaccation. Profiles for pro lenses are much bigger

                   

                  I didn't try lens profile editor yet (I always try to do it 'my way' first  :))

                   

                  However, I find it strange that so much correction is needed on top of automatic correction with pro lens. Prior to lens profiles, I had to use up to about -35 M/G on wide angle with both 400D kit lens and lenses on compacts, decreasing toward almost 0 with rising focal lengths

                  • 6. Re: Tweaking chromatic aberration correction
                    Yammer Level 4

                    Vit Novak wrote:

                     

                    I didn't try lens profile editor yet (I always try to do it 'my way' first  :))

                     

                    However, I find it strange that so much correction is needed on top of automatic correction with pro lens. Prior to lens profiles, I had to use up to about -35 M/G on wide angle with both 400D kit lens and lenses on compacts, decreasing toward almost 0 with rising focal lengths

                     

                    As I understand it, like the DNG Profile Editor, the Lens Profile editor will only work on a "recipe" rather than a "compiled" working profile. It was certainly asking for a *.lcpp file, which I assume is sat on Eric's hard drive

                     

                    I guess either there's a modest variation in lenses, or mine is a bit out, but I always need negative adjustment at any focal length or distance, so an across-the-board offset would be really helpful. Other lenses are less clear cut; my 70-300 zoom does vary a lot in Y/B at different focal lengths. It would be great to have some sort of graphical tool which enables you to re-plot aberration values, although I'd be prepared to work through an XML file, if I knew what to change.

                     

                    I too am surprised that so much manual compensation is required on a lens like this. I have two "gold stripe" lenses which both usually need manual adjustment. I could understand it from cheaper models. In fact my "best" lenses seem to be a 50/1.4 and an "enthusiast" standard zoom (16-85mm). Whether this is down to the optics or the profile, I'm not sure, but I'd be inclined to point at the profiles.

                    • 7. Re: Tweaking chromatic aberration correction
                      Vit Novak Level 3

                      Actually, lens on A650 needs slight positive C/M correction on telephoto end. 400D kit and A620, with less zoom range, don't.

                       

                      In lens profile, item related with this manual CA correction is ScaleFactor under ChromaticRedGreenModel and ChromaticBlueGreenModel (I used only this one to correct CA, although other tags can be used). For example, In case of 18-55 kit lens, I got these values to be ok:

                       

                      FocalLength 18 - ChromaticRedGreen - ScaleFactor = 1.00042, ChromaticBlueGreen - ScaleFactor = 1.00030,

                      FocalLength 35 - ChromaticRedGreen - ScaleFactor = 1.00020, ChromaticBlueGreen - ScaleFactor = 1.00030,

                      FocalLength 55 - ChromaticRedGreen - ScaleFactor = 1.00000, ChromaticBlueGreen - ScaleFactor = 1.00040,

                       

                      It seems that value > 1 means negative C/M and positvive Y/B correction

                       

                      I opened a bunch of photos in PS. Then, Opened a profile with a tex teditor. Changed values, saved profile. Upon switching back to PS, it rereads the file and changes are shown immediately (after second or two delay). So with small profile with 3 points like this, it's relatively easy job

                       

                      I also corrected parameters dealing with distortion and vignttting - it required slightly more effort

                      1 person found this helpful
                      • 8. Re: Tweaking chromatic aberration correction
                        Jeff Schewe Level 5

                        Yammer P wrote:

                         

                        I too am surprised that so much manual compensation is required on a lens like this. I have two "gold stripe" lenses which both usually need manual adjustment. I could understand it from cheaper models. In fact my "best" lenses seem to be a 50/1.4 and an "enthusiast" standard zoom (16-85mm). Whether this is down to the optics or the profile, I'm not sure, but I'd be inclined to point at the profiles.

                         

                        Have you tried simply increasing the Chromatic Aberration correction amount?

                         

                        I've found that on many lens profiles from Adobe (as apposed to the ones done by 3rd parties) that while the CA corrections were going in the right direction, they fell short on the amount of correction on several of my lenses.

                         

                        Check to see if setting the Correction Amount in the Profile to 125-135 amount range. If you find that is sufficient to increase the amount then you can save the correction amount increases in a new "Default" for that lens profile in the Setup drop down menu. Doing this avoids having to add additional manual corrections not associated with the Lens Profile.

                         

                        I also find I'm often selecting a slightly lower amount of correction for the Vignetting correction amount because, well, I actually like a bit of darkening in the corners :~)

                        • 9. Re: Tweaking chromatic aberration correction
                          Yammer Level 4

                          Vit Novak wrote:

                           

                          In lens profile, item related with this manual CA correction is ScaleFactor under ChromaticRedGreenModel and ChromaticBlueGreenModel (I used only this one to correct CA, although other tags can be used). For example, In case of 18-55 kit lens, I got these values to be ok:

                           

                          FocalLength 18 - ChromaticRedGreen - ScaleFactor = 1.00042, ChromaticBlueGreen - ScaleFactor = 1.00030,

                          FocalLength 35 - ChromaticRedGreen - ScaleFactor = 1.00020, ChromaticBlueGreen - ScaleFactor = 1.00030,

                          FocalLength 55 - ChromaticRedGreen - ScaleFactor = 1.00000, ChromaticBlueGreen - ScaleFactor = 1.00040,

                           

                          It seems that value > 1 means negative C/M and positvive Y/B correction

                           

                          I opened a bunch of photos in PS. Then, Opened a profile with a tex teditor. Changed values, saved profile. Upon switching back to PS, it rereads the file and changes are shown immediately (after second or two delay). So with small profile with 3 points like this, it's relatively easy job

                           

                          Many thanks for that, Vit! I'll study one of my larger profiles and see how much of a task it is, and report back with my findings.

                          • 10. Re: Tweaking chromatic aberration correction
                            Vit Novak Level 3

                            This is final version of my 400D kit lens profile - much smaller than original, because I deleted unnecesary tags. Hope it will help.

                            If there are only several points in your profile, it's possible to do it manually. If there are more points, you have option to remove segments dealing with different focus distances for the same focal length - or - find out is there a more elegant way using profile editor ...


                            <?xpacket begin="" id="W5M0MpCehiHzreSzNTczkc9d"?>
                            <x:xmpmeta xmlns:x="adobe:ns:meta/" x:xmptk="Adobe XMP Core 4.4.0-c007 1.126898, 2008/09/29-20:23:32        ">
                               <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
                                  <rdf:Description rdf:about=""
                                        xmlns:photoshop="http://ns.adobe.com/photoshop/1.0/"
                                        xmlns:stCamera="http://ns.adobe.com/photoshop/1.0/camera-profile">
                                     <photoshop:CameraProfiles>
                                        <rdf:Seq>
                                           <rdf:li rdf:parseType="Resource">
                                              <stCamera:ProfileName>Canon EF-S 18-55mm f/3.5-5.6</stCamera:ProfileName>
                                              <stCamera:Make>Canon</stCamera:Make>
                                              <stCamera:Model>Canon EOS 400D</stCamera:Model>
                                              <stCamera:Lens>EF-S18-55mm f/3.5-5.6</stCamera:Lens>
                                              <stCamera:LensID>45</stCamera:LensID>
                                              <stCamera:FocalLength>18.000000</stCamera:FocalLength>
                                              <stCamera:ApertureValue>6.918863</stCamera:ApertureValue>
                                              <stCamera:LensPrettyName>Canon EF-S 18-55mm f/3.5-5.6</stCamera:LensPrettyName>
                                              <stCamera:CameraPrettyName>Canon EOS 400D</stCamera:CameraPrettyName>
                                              <stCamera:CameraRawProfile>True</stCamera:CameraRawProfile>
                                              <stCamera:FocusDistance>5.0</stCamera:FocusDistance>
                                              <stCamera:SensorFormatFactor>1.579033</stCamera:SensorFormatFactor>
                                              <stCamera:PerspectiveModel rdf:parseType="Resource">
                                                 <stCamera:Version>2</stCamera:Version>
                                                 <stCamera:RadialDistortParam1>-0.191200</stCamera:RadialDistortParam1>
                                                 <stCamera:RadialDistortParam2>0.186600</stCamera:RadialDistortParam2>
                                                 <stCamera:ChromaticGreenModel rdf:parseType="Resource">
                                                    <stCamera:RadialDistortParam1>-0.191200</stCamera:RadialDistortParam1>
                                                    <stCamera:RadialDistortParam2>0.186600</stCamera:RadialDistortParam2>
                                                 </stCamera:ChromaticGreenModel>
                                                 <stCamera:ChromaticRedGreenModel rdf:parseType="Resource">
                                                    <stCamera:ScaleFactor>1.00042</stCamera:ScaleFactor>
                                                 </stCamera:ChromaticRedGreenModel>
                                                 <stCamera:ChromaticBlueGreenModel rdf:parseType="Resource">
                                                    <stCamera:ScaleFactor>1.00030</stCamera:ScaleFactor>
                                                 </stCamera:ChromaticBlueGreenModel>
                                                 <stCamera:VignetteModel rdf:parseType="Resource">
                                                    <stCamera:VignetteModelParam1>-0.619983</stCamera:VignetteModelParam1>
                                                    <stCamera:VignetteModelParam2>0.236924</stCamera:VignetteModelParam2>
                                                    <stCamera:VignetteModelParam3>-0.388066</stCamera:VignetteModelParam3>
                                                 </stCamera:VignetteModel>
                                              </stCamera:PerspectiveModel>
                                           </rdf:li>
                                           <rdf:li rdf:parseType="Resource">
                                              <stCamera:ProfileName>Canon EF-S 18-55mm f/3.5-5.6</stCamera:ProfileName>
                                              <stCamera:Make>Canon</stCamera:Make>
                                              <stCamera:Model>Canon EOS 400D</stCamera:Model>
                                              <stCamera:Lens>EF-S18-55mm f/3.5-5.6</stCamera:Lens>
                                              <stCamera:LensID>45</stCamera:LensID>
                                              <stCamera:FocalLength>35.000000</stCamera:FocalLength>
                                              <stCamera:ApertureValue>6.918863</stCamera:ApertureValue>
                                              <stCamera:LensPrettyName>Canon EF-S 18-55mm f/3.5-5.6</stCamera:LensPrettyName>
                                              <stCamera:CameraPrettyName>Canon EOS 400D</stCamera:CameraPrettyName>
                                              <stCamera:CameraRawProfile>True</stCamera:CameraRawProfile>
                                              <stCamera:FocusDistance>5.0</stCamera:FocusDistance>
                                              <stCamera:SensorFormatFactor>1.579033</stCamera:SensorFormatFactor>
                                              <stCamera:PerspectiveModel rdf:parseType="Resource">
                                                 <stCamera:Version>2</stCamera:Version>
                                                 <stCamera:RadialDistortParam1>-0.000600</stCamera:RadialDistortParam1>
                                                 <stCamera:ChromaticGreenModel rdf:parseType="Resource">
                                                    <stCamera:RadialDistortParam1>-0.000600</stCamera:RadialDistortParam1>
                                                 </stCamera:ChromaticGreenModel>
                                                 <stCamera:ChromaticRedGreenModel rdf:parseType="Resource">
                                                    <stCamera:ScaleFactor>1.00020</stCamera:ScaleFactor>
                                                 </stCamera:ChromaticRedGreenModel>
                                                 <stCamera:ChromaticBlueGreenModel rdf:parseType="Resource">
                                                    <stCamera:ScaleFactor>1.00030</stCamera:ScaleFactor>
                                                 </stCamera:ChromaticBlueGreenModel>
                                                 <stCamera:VignetteModel rdf:parseType="Resource">
                                                    <stCamera:VignetteModelParam1>-0.448991</stCamera:VignetteModelParam1>
                                                    <stCamera:VignetteModelParam2>0.844025</stCamera:VignetteModelParam2>
                                                    <stCamera:VignetteModelParam3>-1.906044</stCamera:VignetteModelParam3>
                                                 </stCamera:VignetteModel>
                                              </stCamera:PerspectiveModel>
                                           </rdf:li>
                                           <rdf:li rdf:parseType="Resource">
                                              <stCamera:ProfileName>Canon EF-S 18-55mm f/3.5-5.6</stCamera:ProfileName>
                                              <stCamera:Make>Canon</stCamera:Make>
                                              <stCamera:Model>Canon EOS 400D</stCamera:Model>
                                              <stCamera:Lens>EF-S18-55mm f/3.5-5.6</stCamera:Lens>
                                              <stCamera:LensID>45</stCamera:LensID>
                                              <stCamera:FocalLength>55.000000</stCamera:FocalLength>
                                              <stCamera:ApertureValue>6.918863</stCamera:ApertureValue>
                                              <stCamera:LensPrettyName>Canon EF-S 18-55mm f/3.5-5.6</stCamera:LensPrettyName>
                                              <stCamera:CameraPrettyName>Canon EOS 400D</stCamera:CameraPrettyName>
                                              <stCamera:CameraRawProfile>True</stCamera:CameraRawProfile>
                                              <stCamera:FocusDistance>5.0</stCamera:FocusDistance>
                                              <stCamera:SensorFormatFactor>1.579033</stCamera:SensorFormatFactor>
                                              <stCamera:PerspectiveModel rdf:parseType="Resource">
                                                 <stCamera:Version>2</stCamera:Version>
                                                 <stCamera:RadialDistortParam1>0.080000</stCamera:RadialDistortParam1>
                                                 <stCamera:ChromaticGreenModel rdf:parseType="Resource">
                                                    <stCamera:RadialDistortParam1>0.080000</stCamera:RadialDistortParam1>
                                                 </stCamera:ChromaticGreenModel>
                                                 <stCamera:ChromaticRedGreenModel rdf:parseType="Resource">
                                                    <stCamera:ScaleFactor>1.00000</stCamera:ScaleFactor>
                                                 </stCamera:ChromaticRedGreenModel>
                                                 <stCamera:ChromaticBlueGreenModel rdf:parseType="Resource">
                                                    <stCamera:ScaleFactor>1.00040</stCamera:ScaleFactor>
                                                 </stCamera:ChromaticBlueGreenModel>
                                                 <stCamera:VignetteModel rdf:parseType="Resource">
                                                    <stCamera:VignetteModelParam1>-1.859679</stCamera:VignetteModelParam1>
                                                    <stCamera:VignetteModelParam2>14.866240</stCamera:VignetteModelParam2>
                                                    <stCamera:VignetteModelParam3>-151.614365</stCamera:VignetteModelParam3>
                                                 </stCamera:VignetteModel>
                                              </stCamera:PerspectiveModel>
                                           </rdf:li>
                                        </rdf:Seq>
                                     </photoshop:CameraProfiles>
                                  </rdf:Description>
                               </rdf:RDF>
                            </x:xmpmeta>
                            <?xpacket end="w"?>

                            • 11. Re: Tweaking chromatic aberration correction
                              Yammer Level 4

                              Jeff Schewe wrote:


                              Have you tried simply increasing the Chromatic Aberration correction amount?

                              Yes, it was the first thing I tried. I can't get rid of the CA at any value of Correction Amount. If anything, one R/C or B/Y fringe gets worse as the other improves.

                               

                              Here are some screenshots to illustrate:

                               

                              Uncorrected:
                              ca0.jpg

                              Default correction:
                              ca1.jpg

                              30% extra correction:
                              ca2.jpg

                              100% extra correction:
                              ca3.jpg

                              Default correction + manual correction:
                              ca4.jpg

                              • 12. Re: Tweaking chromatic aberration correction
                                Yammer Level 4

                                Aargh!

                                 

                                Thanks, Vit. I think I'll have to lie down in a darkened room after reading that

                                 

                                You've given me a good idea of how to massage the profile, I just need to bite the bullet now. It's a shame there isn't an easier way to do it, but I'm sure it'll be easier than it looks, once I get stuck in. I just have to make sure any changes are not overwritten at the next update.

                                • 13. Re: Tweaking chromatic aberration correction
                                  Vit Novak Level 3

                                  I took a look to 24-70 lens profile. It's huge. However, there are only 5 different focal lengths, but variuos combinations of focal distances and apertures. Correcting this as is would be very time consuming. If you want to do it manuall and have a kind of automatic version of manual CA correction, you should copy only one  <rdf:li rdf:parseType="Resource"> block for each focal length. You'll end up with 5 blocks. Then, under ChromaticBlueGreenModel and ChromaticRedGreenModel sections delete all tags except ScaleFactor (like I did in above sample). Also, change profile name tag in all block, so PS will show this and original profile. Then, tweak those 10 different scalefactors as I described. Of course, save under different filename first

                                  • 14. Re: Tweaking chromatic aberration correction
                                    Yammer Level 4

                                    Vit,

                                     

                                    What do the other variables stand for?

                                    FocalLengthX/Y

                                    ImageX/YCenter

                                    RadialDistortParam1

                                     

                                    Did you say there was a PDF on the format?

                                    • 15. Re: Tweaking chromatic aberration correction
                                      Yammer Level 4

                                      I've just started experimenting with a profile.

                                       

                                      I duplicated the raw profile and changed both the file name and the profile name, so I get two to choose from.

                                       

                                      I have a photo with focus at infinity, f/4 and 70mm. I adjusted several values of ScaleFactor (eg from 1.000150 to 1.000600) with apertures from f/3 to f/5 at 70mm and the largest distance, but it doesn't seem to affect ACR at all ?!? Am I missing something?

                                      • 16. Re: Tweaking chromatic aberration correction
                                        Vit Novak Level 3

                                        I found it somewhere else, but it's also here

                                         

                                        http://ebookbrowse.com/lensprofile-creator-cameramodel-pdf-d79388582

                                         

                                        All tags are described in the document. As about changes having no affect - I didn't experiment with different focal distances. Is there accurate focal distance information in the raw file? This exif tag has meaningless value in case of 400D kit lens, while in case of compacts it's quite accurate. However, I suggest leaving only one block for each focal length for the beginning, so you end up with only 5 blocks - it will be easier to edit the profile and locate the sorces of problems. I also found some bugs in the ACR regarding this (for instance, when global RadialDistortParam1 is = 1 (I think - not sure), ACR ignores the whole block ... ).

                                        • 17. Re: Tweaking chromatic aberration correction
                                          Yammer Level 4

                                          Brilliant, thanks!

                                           

                                          I'll have a good read and a play with this, and report back later.

                                          • 18. Re: Tweaking chromatic aberration correction
                                            Yammer Level 4

                                            After a bit of playing about, it appears that you only need to change the last set of CA scale values in the LCP file. Maybe each occurrence overrides the previous set? I suppose CA scaling is unlikely to change at different apertures, just the strength.

                                            • 19. Re: Tweaking chromatic aberration correction
                                              Vit Novak Level 3

                                              Didn't try settings for various apertures. I suppose these parameters are interpolated between points specified in the profile (like color matrix between tungsten and daylight), I don't believe they are there to override each other, but I don't know ...

                                               

                                              As you probably know, CA is a combination of transversal and longitudinal. Mentioned Scale parameters correct transverzal component and that one doesn't vary very much with aperture. It's usually red/green. Longitudinal means focus of different wavelengths is falling onto different points (behind or in front of the sensor), usually is manifested as purple fringing and can't be effectively corrected, it can only be masked by decolorizing edges (= defringe option in ACR) and reduced with smaller apertures. Also, it depend where is you focus and what's the distance of various objects in the photo. Worst case scenario with my kit lens is to focus the person several meters from the camera and have some high contrast objects in the background, at wide angle. In that case I'm focusing to something several meters behind the person - this reduces purple fringing significantly

                                              • 20. Re: Tweaking chromatic aberration correction
                                                Noel Carboni Level 7

                                                I could imagine manufacturing differences between copies of the same lens causing the need for a customized profile.

                                                 

                                                Or perhaps even simpler, someone just didn't do a very good job with the original profile as provided by Adobe.  What IS the process for Adobe accumulating lens profile info, anyway?  Is it described somewhere clearly?

                                                 

                                                I am starting to imagine a process by which Camera Raw accumulates information about how much manual correction people dial in over and above the profile, then submits telemetry back to Adobe to help fine tune the profile.  I suppose one would have to assume that manual correction is not being used to ADD distortion on purpose, though.  I suppose that scratches that idea.

                                                 

                                                -Noel

                                                • 21. Re: Tweaking chromatic aberration correction
                                                  Yammer Level 4

                                                  Vit Novak wrote:

                                                   

                                                  Didn't try settings for various apertures. I suppose these parameters are interpolated between points specified in the profile (like color matrix between tungsten and daylight), I don't believe they are there to override each other, but I don't know ...

                                                  The reason I deduced that was because I was working my way through 5 aperture values, starting at f/3, and ending at f/7. The photo was f/4 (the second set), but the picture didn't change until I updated f/7.

                                                  • 22. Re: Tweaking chromatic aberration correction
                                                    Hudechrome Level 2

                                                    Exactly my problem, Noel. I did lens testing for a short time on Tektronix scope camera lenses and I can tell you, it's tedious, especially when assuring identical test conditions when looking over samples separated by a significant time frame (months for sure).

                                                     

                                                    If you look at the link Vit provided, in the description of the data you will find the word "estimated" for certain rgb values. That gave me pause.

                                                     

                                                    Take a look at this site, if you haven't already:

                                                     

                                                    www.slrgear.com

                                                     

                                                    Here's a test of one of my lenses;

                                                     

                                                    http://www.slrgear.com/reviews/showproduct.php/product/97/cat/12

                                                     

                                                    The software they use is provided by DXO, which is why I have confidence in the profiles supplied ib DXO Pro. I can cross check, and it's not hard to validate some of the data like the inflection point in pincushion/barrel vs F/L of zooms. When I checked out initial postings of lenses here, I was agast at the errors, especially in distortion. Further, the substituting of some nearest neighbor lens in Auto when your lens isn't in the list made matters worse.

                                                     

                                                    It's been a long time since I did that kind of testing, but it seems to me that ray tracing still is king in obtaining accurate, repeatable data. I don't understand how substituting a single ray centered on the object/perspective center thru the image plane can be anything more than a stating point, and when it includes "estimated" in the data collection, well, maybe I forgot too much!

                                                     

                                                    Imo, anyway!

                                                    • 23. Re: Tweaking chromatic aberration correction
                                                      Hudechrome Level 2

                                                      I don't know about you, Yammer, but many times my deduces are wild!

                                                      • 24. Re: Tweaking chromatic aberration correction
                                                        Yammer Level 4

                                                        I must admit, it didn't make sense to me. But why would it only affect f/4 CA when changing f/3, f/4, f/5, and f/6 data had no effect?

                                                        • 25. Re: Tweaking chromatic aberration correction
                                                          Hudechrome Level 2

                                                          Those are questions needing exploration.

                                                          • 26. Re: Tweaking chromatic aberration correction
                                                            Vit Novak Level 3

                                                            Well, there are probably still some bugs in the code. I faced at least one bug and also managed to crash ACR with wrong parameters, sometimes it shown message that profile couldn't be read (when parameters were out of the range) etc ... I suppose whole thing is meant to work in connection with lens profile generator instead of making these profiles manually. And I must admit I didn't read whole documentation

                                                             

                                                            Still suggesting to do it the same way as I did - Just forget about various apertures and distances when doing this manually, leave just one block for each focal length and some medium aperture and subject distance - it is proved to work and it corrects say 95% of tranversal CA on most photos with my cheap-plastic-fantastic lenses - don't care about that 5%. On the other side, Jeff admited that their profiles correct 70% of CA, as you have to apply 130% correction with his lenses ...so there is still some room for improvement ... like it was with color management in ACR (still waiting Eric to address banding issue I documented a while ago) ... but with some manual corrections, these lens profiles are perfectly usable even now, at least for me ...

                                                            • 27. Re: Tweaking chromatic aberration correction
                                                              Yammer Level 4

                                                              I was hoping Eric might chime in on this one.

                                                              • 28. Re: Tweaking chromatic aberration correction
                                                                Hudechrome Level 2

                                                                I believe Adobe's method to date in handling aberrations (Noise, CA, Distortion, sharpness) is fatally flawed. If we did color management from the same perspective (Here's a bunch of sliders, tweak them until things look ok) we would be in a world of hurt. The factors I mentioned are far more subtle (what is information, what is noise) than color deviation, but nonetheless equally important.

                                                                 

                                                                I have never, never encountered images at 100% equivalent magnification under an enlarger with a magnifier that even suggests that this kind of approach for the user is at all necessary. The closest ever was with Tech Pan where you could actually oversharpen with the wrong development procedure to produce the exaggerated edge borders that software sharpening can produce. While it wasn't a simple matter to optimize development, it was straight forward enough to get it done once and go on with your photographic life.

                                                                 

                                                                Today we have inferior optical performance due to the sensors and the lazy attitude towards producing lenses that actually perform right out of the box. Fix it with software. OK, fine, but fix it at the source, with known test targets and do it behind the users , not in front of them. Get it done and leave it alone. You cannot solve a signal to noise ratio problem by tweaking it by eye or ear and call it any kind of standard, period.

                                                                 

                                                                I completely embrace digital photography. I do not (well most of the time!) look back wistfully. But having to relearn lessons learned in the past which are solvable,engineering problems sometimes put me in a state of despair over current engineering practices.

                                                                 

                                                                Here's my bottom line:

                                                                 

                                                                I don't want to deal with sensor noise problems

                                                                I don't want to deal with CA and other optical aberrations

                                                                I don't want to deal with system sharpness errors

                                                                I don't want to deal with distortion

                                                                I do want to be able to introduce my gear into the app and have it do the above with one click, and have the parameters available, should I decide otherwise. Then I deviate from a known, valid correction in the first place.

                                                                 

                                                                Hopefully, the hardware manufacturers will see the wisdom in providing this as standard corrections applied to the RAW file itself, as default.

                                                                • 29. Re: Tweaking chromatic aberration correction
                                                                  Noel Carboni Level 7

                                                                  Hudechrome wrote:

                                                                   

                                                                  I believe Adobe's method to date in handling aberrations (Noise, CA, Distortion, sharpness) is fatally flawed. If we did color management from the same perspective (Here's a bunch of sliders, tweak them until things look ok) we would be in a world of hurt.

                                                                   

                                                                  I don't see this as a flawed method, so much as flawed lens profiles, or an inability to deal with specific lens copies.

                                                                   

                                                                  For what it's worth, the packaged profiles for the lenses I use work quite well, with only a very small amount of tweaking (if at all) needed by me.  Usually it's a situation where I'm being very picky and the CA in one corner of my image is corrected while the CA in another corner is slightly off (which I attribute to manufacturing tolerances in the lens, not a problem with the software).

                                                                   

                                                                  In EVERY case the results are better than I'd get with shooting onto film.  Any remnant distortions or aberrations are beyond the current technology (see next year's ACR release).  For example, is there any reason profile-guided deconvolution couldn't be used to correct "smearing" near the corners of an image?

                                                                   

                                                                  I don't want to deal with sensor noise problems

                                                                  I don't want to deal with CA and other optical aberrations

                                                                  I don't want to deal with system sharpness errors

                                                                  I don't want to deal with distortion

                                                                   

                                                                  Just because you don't want to fine tune the various aberration corrections doesn't mean others don't.  I for one am very happy to have the ability to do so - there's not always just one solution (e.g., do you want the entire FOV or do you want straight lines with certain wide angle lenses?).

                                                                   

                                                                  You have what you need in ACR to just set up defaults and just use them without thinking any more about it, so where's the problem?

                                                                   

                                                                  -Noel

                                                                  • 30. Re: Tweaking chromatic aberration correction
                                                                    Hudechrome Level 2

                                                                    But it still isn't "One stop Shopping" so to speak.

                                                                     

                                                                    I went through some recent shots and turned on the corrections in ACR. So far as it goes, I see reasonable corrections. Not long ago, I did this test and layered DXO over ACR, limiting DXO to only the corrections available in ACR. The two files nulled nicely which was far from true a year or so ago. But, I still have  the rest of the corrections I can obtain directly in DXO to correct in ACR, so I am left with either hand corrections or reverting to DXO.

                                                                     

                                                                    Silly way to run a correction suite, IMO.

                                                                     

                                                                    When I said what I don't want to do, please understand that refers to default corrections. As you say, each individual unit may not fit the profile exactly and you have to go in. That's fine. I already know what I have to do for my stuff to finesse DXO defaults.

                                                                     

                                                                    Why I said fatal flaw is that there seems to be no concerted effort to make it complete, and even worse, using profiles generated by users exacerbates

                                                                    the nature of fatal flaw. How do we know the condition of the DUT? There isn't any control whatsoever.

                                                                     

                                                                    Some folks love to do the software writing. More power to them! I read as many as I can, both here and elsewhere.

                                                                     

                                                                    I believe the answer lies with the OEM's. They need to provide the data, perhaps as DXO does, as a DNG perhaps, which PS can handle. Otherwise, it's a big job with providers duplicating each others efforts. According to my info, DXO lens tests run 30 hrs on the average to complete and validate their profiling.

                                                                    • 31. Re: Tweaking chromatic aberration correction
                                                                      Noel Carboni Level 7

                                                                      In all that I believe you said these two things:

                                                                       

                                                                      • You think it's error-prone to have the user base submit profiles (I happen to agree, though it's an intriguing idea).
                                                                      • You feel DXO Optics Pro does more sophisticated stuff than Camera Raw in terms of correcting lens flaws.

                                                                       

                                                                      Please correct me if I got that wrong or left anything out.

                                                                       

                                                                      -Noel

                                                                      • 32. Re: Tweaking chromatic aberration correction
                                                                        Hudechrome Level 2

                                                                        Correct on both.

                                                                         

                                                                        Understand that DXO is in the business of providing software for optical applications. The website www.slrgear.com publishes pretty nifty results of their testing of camera optics. The tool they use (at least last time I checked, about a year ago) is a dxo optics package.

                                                                         

                                                                        Here is another DXO site worth visiting:

                                                                         

                                                                        http://www.dxomark.com/index.php

                                                                         

                                                                        So yes, sophisticated. But they cannot hold a candle to PS concerning the rest of image editing. There is no comparison, and their B&W conversion package is woefully inadequate. Nik has a far better B&W than DXO and in many ways, better than Adobe. But that one is in the eyes of the beholder.

                                                                         

                                                                        I gravitate to DXO on these matters because, from my pov, their approach is basically objective, whereas Adobe is subjective, whcih isn't surprising as image editing is a subjective process. I would be far more comfortable if Adobe took optical corrections completely out of the subjective realm. It's absurd to do otherwise. Pursuing the subjective mode is why I said fatally flawed.

                                                                         

                                                                        Why did you say "...though it's an intriguing idea".?

                                                                        • 33. Re: Tweaking chromatic aberration correction
                                                                          Yammer Level 4

                                                                          Hudechrome wrote:

                                                                           

                                                                          I believe Adobe's method to date in handling aberrations (Noise, CA, Distortion, sharpness) is fatally flawed.

                                                                          Let's not get ahead of ourselves. I'm inclined to think that ACR is not behaving as designed, and that there is quite possibly a bug or two which needs fixing. Why? I just demonstrated something which makes no sense...

                                                                           

                                                                          If you were going to design a system which corrected CA using reference data from sample points, you'd expect the software to interpolate between samples, wouldn't you?

                                                                           

                                                                          So, in the above case, you would expect a photo taken at 70mm, f/4, and towards infinity to be using data points from the 70mm set, and this seems to be the case.

                                                                          Next, you'd expect it would use data from the furthest focal distance, and this also appears to be the case.

                                                                          Finally, you'd expect it to interpolate data between f/3 and f/5, or even use the f/4 value on its own. This doesn't appear to be the case. In fact, only manually editing the f/7 data (last in the correction profile) seems to have an effect. To me, this looks suspiciously like a bug. But without knowing how the software works under the bonnet, I can only speculate.

                                                                           

                                                                          If I am correct, I'd be a lot more impressed with Adobe's system, and I am happy to accept whatever methodology they choose, besides, at the end of the day, I think that photography is subjective anyway.

                                                                          • 34. Re: Tweaking chromatic aberration correction
                                                                            Vit Novak Level 3

                                                                            Yes, that would be expected behaviour, although it is not defined in mentioned pdf document

                                                                            • 35. Re: Tweaking chromatic aberration correction
                                                                              Yammer Level 4

                                                                              What is? What I'd expect, or what actually happens?

                                                                              • 36. Re: Tweaking chromatic aberration correction
                                                                                Vit Novak Level 3

                                                                                I mean, I would expect the same behaviour as you

                                                                                 

                                                                                Tried this with my profile for 400D kit lens. Duplicated last block (for focal length 55 mm), then in first block I changed aperture to 5.6, in second block I changed it to 22. Opened a photo taken with focal length 55mm and aperture 6.3. In this case, changing scale parameter in both blocks had effect - ACR seems to be interpolating effective scale factor for actual aperture based on those two points

                                                                                 

                                                                                But, after digging more, it turned out that this interpolation doesn't work exactly as I expected. I tried this: changed scale factor of ChromaticRedGreenModel for first block (aperture 5.6) to as high as 1.2, and changed aperture of second block to 6.3. As expected, high scale factor of first block didn't have any effect, because result of interpolation was scale factor from the second block. Then, started raising aperture for the second block. Nothing happened until I raised it to 6.92, when scale factor of the first block started "kicking in". Strange ... why this happened at 6.92 and not at 6.31 or 6.32 ? I'm not saying it's a bug, maybe it is made this way by design, but ...

                                                                                 

                                                                                Also, I crashed ACR several times with scale factor being way too high or too low ...

                                                                                 

                                                                                Eric, some comment about this ?

                                                                                • 37. Re: Tweaking chromatic aberration correction
                                                                                  Hudechrome Level 2

                                                                                  Photography (for the most part) is subjective, optics, not. Neither is color management. CM is where Adobe does a better job, and takes subjectivity out of the loop.

                                                                                   

                                                                                  I'm not suggesting Adobe abandon the manuals at all. I  strongly support starting with a basically corrected image, for which one can further do what one wishes, creative sharpening, as it is called.

                                                                                   

                                                                                  Compared to analog photography, we are presented, out of the camera, a crippled image. Even the early images were far better technically than what passes for a digital image before fixing (Look at a Daguerreotype) It has to be fixed before anything else. However, look at what some of the stock agencies want. They want unsharpened images! (or did. Do they still demand unsharpened?)  Now, why would that be?

                                                                                   

                                                                                  Supposing you could not obtain a lens, back in those early days, which was corrected across the board, or rather, was left less than optimum, but with a series of corrective optics which you as the user, had to employ as you photographed. And suppose you could not know what was the best combination of glass to have employed until you looked at the negative. And suppose several of you were looking at the corrections necessary and disagreed as to what is optimum? What would you do?

                                                                                   

                                                                                  Sometimes, I feel more like a sculptor than a photographer. I have to chip away at the marble containing my "masterpiece" which at least, I can see in more than my mind's eye, just to clean it up!  Or, I can apply a series of algorithms that also "see" my "masterpiece" and know what to do to bring in into alignment with known, proven output that reveals it in it's correct state, with only one tap on the chisel.

                                                                                   

                                                                                  I suppose people are wincing over my use of terms like "fatally flawed" or "crippled". The Emperor isn't wearing clothes, folks, or at least, isn't properly dressed!

                                                                                  • 38. Re: Tweaking chromatic aberration correction
                                                                                    Noel Carboni Level 7

                                                                                    Hudechrome wrote:

                                                                                     

                                                                                    Why did you say "...though it's an intriguing idea".?

                                                                                     

                                                                                    From a software developer's perspective, consider this:

                                                                                     

                                                                                    What if you could provide a software package that's web-connected, and accumulates useful information all by itself from the huge batch of users out there via telemetry.  Voila, you don't have to spend money to research all the available lenses and you don't have to do all the testing yourself before releasing the product!

                                                                                     

                                                                                    If you can create some kind of closed-loop system that tunes itself, and you haven't screwed anything up too badly, sooner or later your database will be self-populated with data from the masses that is actually the best average of all the lens information available in the world.

                                                                                     

                                                                                    There are two fundamental assumptions with this approach, however:

                                                                                     

                                                                                    • The masses will correctly judge lens characteristics properly for the most part.
                                                                                    • You, as the developer, haven't screwed anything up causing bogus results.

                                                                                     

                                                                                    The payoff can be huge!  You can create a system that accumulates information about the entire market almost immediately, without big expenditure.  Only thing is it may not be perfectly accurate right away.  But it will correct itself in time.

                                                                                     

                                                                                    Maybe you augment the database with some test data you accumulate yourself via a modestly-funded test lab to help things along a bit.

                                                                                     

                                                                                    Like I said, from a developer's perspective, a VERY intriguing idea.

                                                                                     

                                                                                    -Noel

                                                                                    • 39. Re: Tweaking chromatic aberration correction
                                                                                      Hudechrome Level 2

                                                                                      You would, of course, have to have a "Gold Standard", that is an image which meets the theoretical limits as close as possible, that is error decreases to the vanishing point. Then collect your data for that lens type on that specific sensor and see how it averages out.

                                                                                       

                                                                                      I wince at all that. I don't want average output. I want premium output.

                                                                                       

                                                                                      BTW, in order to collect my data, you will have to pay for it. Intellectual property rights and all that!

                                                                                       

                                                                                      I've been studying ACR vs DXO again today, running shots from a garden setup for sharpness and noise, which included stuff like tripod vs VR as well. It appears that DXO varies the noise reduction as some function of density; more NR in darker areas for a given ISO. Or maybe it is simply more noticeable in those areas.

                                                                                       

                                                                                      It sticks in my craw that I have to take my images and begin by discarding data, just to get sharpness. Where does the degradation begin? If I put a Leica 50mm f2 Summicron  on a high end full size sensor and compare images obtained first on Tech Pan then the sensor, will the image be inherently softer off the sensor? The Summicron comes as close to the theoretical limits of optical performance as any optic ever, so I found from an independent study of such optics. If that is so, I should not have to do anything at all to sharpen that image, at ordinary contrast values.

                                                                                      1 2 3 Previous Next