19 Replies Latest reply: Jul 9, 2012 6:38 AM by mschlotz RSS

    [LR 4.1] Graduated filter bug

    Samoreen Community Member

      Hi,

       

      I have an image horizontally divided in two parts : 2/3 (sky) and 1/3 (ground). I have created 2 graduated filters, each fully and exactly covering one part of the image. I created the biggest one (2/3) first. The two graduated filters do not overlap.

       

      When I apply effects to the first graduated filter, everything happens as expected : any change made to the exposure, saturation or whatever parameter only affects the area covered by the graduated filter.

       

      When I (try to) apply effects to the second graduated filter (1/3 - covering the rest of the image), any change is applied to the whole image, as if the graduated filter was covering all the image.

       

      It's too severe a bug to imagine that nobody noticed it until now. On the other hand, this behavior is not logical.

       

      It's not specific to a particular image. I can reproduce this on any image.

       

      Did I miss something obvious?

       

      Anyone already seen this ?

       

      Thanks in advance.

        • 1. Re: [LR 4.1] Graduated filter bug
          retiredff Community Member

          Not often that I have to place two on the same photo, but I noticed the same thing a while back. I changed either the placement or the direction of drag and it corrected the problem. Don't have time to play this morning, but try either placing both filters so they are each dragged down, or place the bottom one so it has to be dragged up. Wish I could remember which one I did, but it was a simple fix for me.

          • 2. Re: [LR 4.1] Graduated filter bug
            Hal P Anderson Community Member

            Not a bug. The filters are working as designed. The space between the lines is where the gradient happens. The space above the top line gets the full impact of the change, and everything below the bottom line gets none. With your smaller gradient, the full change gets done to everything above it, which includes the sky, hence your problem.

             

            Try flipping the smaller gradient upside down, so the bottom line goes to the top.

             

            Hal

            • 3. Re: [LR 4.1] Graduated filter bug
              Samoreen Community Member

              Thanks retiredff,

               

              Thanks for the tips but this is not a workable solution : for example, creating a filter by dragging up and applying a negative value to  the saturation slider is not the same as creating an equivalent filter by dragging down and applying a positive value to the saturation slider.

               

              I could actually determine when the problem occurs and I guess anyone can reproduce : if you create a filter by dragging down and if the anchor point is located in the lower half of the image, you'll have the problem. Ditto when creating a filter by dragging right if the anchor point is located in the right half of the image.

               

              Creating a single or multiple graduated filters is not related to this problem.

              • 4. Re: [LR 4.1] Graduated filter bug
                Samoreen Community Member

                Hi,

                Hal P Anderson wrote:

                 

                Try flipping the smaller gradient upside down, so the bottom line goes to the top.

                 

                 

                So what if I want to apply say, a low saturation from a point located below the middle of the image, progressively returning to the normal saturation at the bottom of the image, without affecting the upper part of this image?

                 

                Thanks.

                • 5. Re: [LR 4.1] Graduated filter bug
                  Samoreen Community Member

                  Samoreen wrote:

                   

                  So what if I want to apply say, a low saturation from a point located below the middle of the image, progressively returning to the normal saturation at the bottom of the image, without affecting the upper part of this image?

                   

                  Maybe a "Reverse effect direction" check box is missing ?

                  • 6. Re: [LR 4.1] Graduated filter bug
                    mschlotz Community Member

                    By any chance did you not hit the "NEW" icon before doing the second grad filter?

                    • 7. Re: [LR 4.1] Graduated filter bug
                      Samoreen Community Member

                      mschlotz wrote:

                       

                      By any chance did you not hit the "NEW" icon before doing the second grad filter?

                       

                      I did. Otherwise, I wouldn't get 2 anchor points. But as mentioned above, the problem is actually not the number of filters.

                      • 8. Re: [LR 4.1] Graduated filter bug
                        trshaner CommunityMVP

                        Samoreen wrote:

                         

                         

                        So what if I want to apply say, a low saturation from a point located below the middle of the image, progressively returning to the normal saturation at the bottom of the image, without affecting the upper part of this image?

                         

                        Thanks.

                        !) Create two overlapping Gradient Filters dragged from the top-down, with one slightly below the other.

                         

                        2) Set the upper filter to + Saturation (+100) and  the bottom filter to an equal amount - Saturation (-100) value.

                         

                        3) It may take a bit of repositioning to get the desired effect, but it will work just fine and not disturb the top part of the image.

                         

                        Message was edited by: trshaner

                        • 9. Re: [LR 4.1] Graduated filter bug
                          DavidRitch Community Member

                          I tried to reproduce this.  I'm using LR 4.1 under Windows 7 64-bit.

                           

                          I opened the Graduated Filter Tool, and clicked in the middle of the image.  I dragged down about half way to the bottom of the image.  This created a graduated filter.  I played with graduated filter's exposure adjustment.  I found that the adjustments took full effect in the top half of the image - above the top line of the graduated filter.  It had no effect below the bottom line of the filter.  The effect varied smoothly in intensity between the top line and bottom line of the graduated filter.

                           

                          This is the normal and expected behavior.  I believe that you are saying that when you do this, the effect is identical across the whole image, regardless of the the 3 lines associated with the filter.  Is that correct?  If so, that is certainly not the expected behavior, and sounds like a bug.

                           

                          David

                          • 10. Re: [LR 4.1] Graduated filter bug
                            Samoreen Community Member

                            Thanks trshaner,

                             

                            I'll try this.

                            • 11. Re: [LR 4.1] Graduated filter bug
                              Samoreen Community Member

                              DavidRitch wrote:

                               

                              This is the normal and expected behavior.  I believe that you are saying that when you do this, the effect is identical across the whole image, regardless of the the 3 lines associated with the filter.  Is that correct?

                               

                              Actually, since the second filter was covering all the bottom part of the image, I perceived the changes (maximum effect) as applying to the upper part while I did not expected anything to change in that area (because I made other settings with another filter). I wrongly interpreted this as "the whole image".

                               

                              So while I admit that it's not a bug "per se", I realize that something is missing : the ability to apply the graduated effect in the reverse direction. Which is not the same as flipping the filter upside down.

                               

                              The position from which the dragging begins is the maximum effect location. When dragging downwards, this means that everything above this line will be affected by the "maximum effect" as well. Which was unwanted in my case. If I could start dragging upwards from the bottom of the target area and say "please apply the effect in the reversed direction", the problem would be solved because the upper area of the image would not be affected.

                              • 12. Re: [LR 4.1] Graduated filter bug
                                LRuser24 Community Member

                                So while I admit that it's not a bug "per se", I realize that something is missing : the ability to apply the graduated effect in the reverse direction. Which is not the same as flipping the filter upside down.

                                Either I am missing something, or flipping the filter upside down (or dragging upwards from the start) is exactly what you need - you only have to apply the inverse value of the effect you want to use.

                                 

                                Example: You want to make the lower half of the image <i>brighter</i> (as opposed to the common case to make the upper half [sky] darker): Set exposure to a <i>positive</i> value and then drag upwards.

                                 

                                But perhaps I didn't understand your point correctly...

                                 

                                EDIT: Ok, perhaps now I understood: I looked again at your question: "So what if I want to apply say, a low saturation from a point located below the middle of the image, progressively returning to the normal saturation at the bottom of the image, without affecting the upper part of this image?": So you want to "reverse" the filter effect in such a way that it:

                                • immediately (abruptly) starts at full strenght at line #1
                                • gradually dropping to zero strengts at line #2

                                Is that correct? If it is: No, that is not directly possible. Probably because the graduated filter are modelled after real-world filters, which cannot do such things (AFAIK). You have to use two filters together to get the desired effect, just like trshaner described above. Here is a quickly done example - hopefully this is exactly what you like to achive:

                                 

                                20090521-IMG_0309.jpg

                                For this example, I used a very thin GF (dragged upwards) at the line where the saturation sharply drops (set to -100) and a larger GF (also dragged upwards) spanning from the bottom of the picture to the first GF, with saturation +100, so it gradually counteracts the first GF.

                                • 13. Re: [LR 4.1] Graduated filter bug
                                  martin-s Community Member

                                  It's often much easier to use a brush instead. A large brush with a maximum feathered edge is a graduated filter that isn't limited to a straight line. Once it's in place you can erase it on one side using a much less feathered edge giving you an effect that's impossible to achieve with a graduated filter, example below.

                                   

                                  There are endless possibilities using this technique.

                                   

                                  Lightroom001.jpg

                                  • 14. Re: [LR 4.1] Graduated filter bug
                                    Samoreen Community Member

                                    Thanks, LRuser24.

                                    LRuser24 wrote:

                                    EDIT: Ok, perhaps now I understood: I looked again at your question: "So what if I want to apply say, a low saturation from a point located below the middle of the image, progressively returning to the normal saturation at the bottom of the image, without affecting the upper part of this image?": So you want to "reverse" the filter effect in such a way that it:

                                    • immediately (abruptly) starts at full strenght at line #1
                                    • gradually dropping to zero strengts at line #2

                                    Is that correct?

                                     

                                    If the filter was created by dragging upwards and if line #1 is above line "2, yes your statement is correct.

                                     

                                    LRuser24 wrote:

                                     

                                    If it is: No, that is not directly possible. Probably because the graduated filter are modelled after real-world filters, which cannot do such things (AFAIK).

                                     

                                    Some real world filters can do this : reverse neutral density filters, for example. The video available here shows how to simulate such a filter in Lightroom by using a technique similar to what has been suggested above. Unfortunately, this doesn't work with all parameters. It works with exposure which is truly "reversible" but not for saturation.

                                     

                                    So I'd like the graduated filters to be able to optionally work like those reverse neutral density filters but for all parameters.

                                     

                                    Anyway thank you all for the advice and suggested workarounds.

                                    • 15. Re: [LR 4.1] Graduated filter bug
                                      Samoreen Community Member

                                      Thanks, martin-s. I'll give a try to this approach.

                                      • 16. Re: [LR 4.1] Graduated filter bug
                                        Samoreen Community Member

                                        Samoreen wrote:

                                         

                                        It works with exposure which is truly "reversible" but not for saturation.

                                         

                                        I need to be more accurate here. For some parameters like saturation, clarity, sharpness, etc. I'm not sure about the meaning of +100 and -100 in a local adjustment. This is why I think these parameters are not "reversible". What such a setting does depends heavily on the "starting point" of the filter. So I'm not sure that +100 will exactly compensate for -100. For example, will the result depend on the filter creation order? I need to experiment more with the approach that you and trshaner suggested.

                                         

                                        But I think that a "reverse direction" option would be safer and a more logical answer to this issue. However, I have no hope to see this appear in LR4. Not many users will need this and I think that the problem can be solved easily (if not quickly) in Photoshop. Anyway, there are more severe problems in LR waiting for a fix since version 1... .

                                        • 17. Re: [LR 4.1] Graduated filter bug
                                          LRuser24 Community Member

                                          Some real world filters can do this : reverse neutral density filters, for example. The video available here shows how to simulate such a filter in Lightroom by using a technique similar to what has been suggested above. Unfortunately, this doesn't work with all parameters. It works with exposure which is truly "reversible" but not for saturation.

                                           

                                          So I'd like the graduated filters to be able to optionally work like those reverse neutral density filters but for all parameters.

                                          I didn't know about these reverse neutral density filters - thanks for the video link.

                                           

                                          But I still don't understand why you think that it is not possible using saturation. My example shows that it is possible - and it is possible with any other paramter. EDIT: You posted your clarification just before I posted my reply - now I understand what you mean. From my experience, all paramter +/- values cancel themselves out perfectly (and it does not depend on the creation order), but you are right, a "native" support by LR would be safer and easier than using two GFs.

                                           

                                          P.S. The suggestion of martin-s is of course also a good alternative.

                                          • 18. Re: [LR 4.1] Graduated filter bug
                                            Samoreen Community Member

                                            LRuser24 wrote:

                                            My example shows that it is possible - and it is possible with any other paramter.

                                             

                                            I don't say it's not possible. Your example works, that is, it shows that something can be done to get a result that is nearing what I want. It's probably an acceptable workaround.  But I'm wondering whether the result is the same as if I were applying a reverse graduated filter directly. I guess it's not the case.

                                            • 19. Re: [LR 4.1] Graduated filter bug
                                              mschlotz Community Member

                                              oops, I miss read your note, you are correct... this issue "bug?" is a bit odd