10 Replies Latest reply on Mar 12, 2009 2:45 PM by (Jeff_Schewe)

    Canon 5D2 ACR chroma/luminance noise reduction

    pdcompton Level 1
      Admitting to being somewhat lazy w/respect to experimentation, can anyone recommend chroma/luminance noise reduction settings in ACR for the 5D2, as a function of ISO?
        • 1. Re: Canon 5D2 ACR chroma/luminance noise reduction
          Panoholic Level 2
          This question exposes a basic misunderstanding of the nature of the connection between ISO and noise.

          The following paper on this issue has been deemed too steep and too pixel peeping by some wievers and I plan to rework it, but that takes some more time:

          http://www.cryptobola.com/PhotoBola/SourceOfNoise.htm
          • 2. Re: Canon 5D2 ACR chroma/luminance noise reduction
            pdcompton Level 1
            Thanks, but my question was somewhat more basic. I just purchased a 5D2 and have yet to 'develop' any raw images. I was hoping to leverage the experience of others on what ACR chroma and luminance noise reduction parameters they found to work well.
            • 3. Re: Canon 5D2 ACR chroma/luminance noise reduction
              Level 1
              > I just purchased a 5D2 and have yet to 'develop' any raw images.

              Shoot some, test out some settings and get back to us...it would be really silly to simply take what somebody else gives you without doing your own tests...
              • 4. Re: Canon 5D2 ACR chroma/luminance noise reduction
                Panoholic Level 2
                > I was hoping to leverage the experience of others on what ACR chroma and luminance noise reduction parameters they found to work well

                Had you read the linked paper, you would perhaps understand, that this is not the question of ISO primarily but of exposure; that is the whole point. Furthermore, you have to understand your *raw* data (not what ACR is displaying), to "get behind the noise".

                Example: the noisy blue sky. If you look at the raw channels, you notice, that the red level is very low even in an area, which looks well-exposed, and *that* causes the noise.

                THink about this: whenever I see a post with someone boosting how great ISO 1600 of his camera is, showing a well-exposed shot, or the opposite, complaning about high noise at ISO 100 and it is obvious, that the shot was underexposed, I know that the poster does not understand the correlation betwen exposure, ISO and noise.
                • 5. Re: Canon 5D2 ACR chroma/luminance noise reduction
                  Ramón G Castañeda Level 4
                  Thank you once again, Gabor. Nice paper. ;)

                  For those too lazy to grapple with the whole paper, at least read the conclusions at the end:

                  Conclusions

                  * The source of noise is not the high ISO setting but low exposure.

                  * Higher ISO makes up for some of the noise caused by low exposure.

                  * Increasing the ISO setting decreases the noise up to a certain ISO level; further increasing the ISO setting is practically useless for the raw image.

                  * The noise reducing effect of one ISO stop is the highest in the low ISO range, diminishing in the higher ISO range.

                  * The higher the ISO, the lower the dynamic range. This is not obvious from the above demonstrations, it needs some explanation. Increasing the ISO increases the numerical pixel values. Those values are limited by the bit depth and other miraculous circumstances; thus, increasing the ISO by one stop "pushes out" the very top stop of the highlights. This decreases the dynamic range at the high end by one stop. On the other hand, the dynamic range is now greater at the dark end due to the decreased noise. For example if the noise reduction by the higher ISO is 2/3 EV, then the loss of the dynamic range is 1/3 EV.
                  • 6. Re: Canon 5D2 ACR chroma/luminance noise reduction
                    Ramón G Castañeda Level 4
                    Above all, think about the implications of Jeff's post #3. Buying a camera of that caliber and looking to use someone else's settings indiscriminately just boggles the mind.
                    • 7. Re: Canon 5D2 ACR chroma/luminance noise reduction
                      pdcompton Level 1
                      Jeff - Thanks much and I intend to do just that. Having had the camera a short time, I have not had time to experiment and was curious of other's experience so I could add to my tests.

                      Gabor - You are right, I did not understand well the correlation between exposure, ISO and noise. Your paper has helped immensely in understanding why what I thought were decent exposures were sometimes disappointing.

                      Ramon - I don't recall indicating I would simply use someone's settings indiscrimantly; those are your words. I was just curious as to what worked and what did not so I could try the same and judge the results. But, hey this is the internet where you can insult and make condescending remarks more or less anonymously.
                      • 8. Re: Canon 5D2 ACR chroma/luminance noise reduction
                        Ramón G Castañeda Level 4
                        >those are your words.

                        My words and your typos, not mine. :D
                        • 9. Re: Canon 5D2 ACR chroma/luminance noise reduction
                          Panoholic Level 2
                          Well, if we deal with the subject here at this level of details, then I need to add some points:

                          1. For the advanced digital photographer: if a lower ISO is enough for a correct exposure, but one is aiming at achieving "exposure to the right", then it is useful to turn up the ISO
                          i without reducing the exposure
                          in order to "get to the right edge".

                          2. Mitigating the above: the vast majority of cameras do not have true 1/3 stop ISO steps, i.e. there is no analog gain associated with the 1/3 steps; they are achieved by numeric manipulation of the nearest (lower or higher) full stop ISO result. For example from the Canons, only the 1Dxxx models support real 1/3 step ISOs.

                          There is no point of using these ISO steps with raw data.

                          3. Almost all cameras offer high ISOs, which are fake, i.e. numerical derivations of lower ISOs; in some cases they are
                          b not only
                          those characterized as "High", "Extention", etc. For example ISO 12800 and 25600 are marked as "high" with the Canon 5DMkII, but in reality already 6400 is fake.

                          There is no point of using these ISO steps with raw data (yes, I wrote this already).

                          4. The top
                          b useful
                          ISO is usually even lower. For example the graphs in http://www.panopeeper.com/Demo/Canon5DMkII_Noise.GIF show, that the loss from 1600 to 3200 with the 5DMkII is precisely one stop, i.e. there is no point to use ISO 3200 with raw data. Some other cameras can not utilize even lower ISO settings in raw (the usefulness of those settings is when recording JPEG in-camera).

                          Using those ineffective ISO steps causes cutting off the one stop of the dynamic range with each ISO stop.

                          This is an ACR forum, thus these issues are off-topic, but so much can be perhaps tolerated.
                          • 10. Re: Canon 5D2 ACR chroma/luminance noise reduction
                            Level 1
                            >This is an ACR forum, thus these issues are off-topic, but so much can be perhaps tolerated.

                            Not at all...I think anything related to image quality in raw captures is useful territory in the Camera Raw forum...

                            I think camera noise and ISO are fruitful topics!