Jump to content

very bad purple fringing with 50D how about 7D?


douwe_spoelstra

Recommended Posts

<p>Magnus,</p>

<p>If you look hard enough you should see two coloured chromatic aberration on any image, particularly the corners, especially if the lens is not apochromatically corrected.</p>

<p>That is one of the areas where digital really kills itself, because we can all zoom into 100% we do, we never printed the majority of our images anywhere near that size so didn't notice how "bad" our lenses are. They are the same lenses though.</p>

Link to comment
Share on other sites

<p>Brett - I'ved used that as a 'quick fix' for numerous photos in the past, usually when there is a little CA at the corners and I can't be bothered to go back through ACR to adjust. No problems here. I print fairly big too, although I'm never one to pixel peep. Although I'm always one to find a solution, rather than whinge to strangers over the internet ;)</p>
Link to comment
Share on other sites

<p>I cannot answer the 50D v. 7D question. But I can tell you that the problem occurs on all Canon cropped sensor and full frame DSLRs that I have used in extreme exposure situations like the one you have shown, and that the solutions include the steps I mentioned in my earlier post.</p>

<p>I cannot rule out the possibility that the 7D is somehow "worse" than other cropped sensor bodies like the 50D since I have not used either of those specific cameras. I can speculate that a) this is a type of artifact that is common to DSLRs due to the way they deal with over exposure, b) that to the extent that it is a sensor related issue it <em>may</em> appear to be "larger" in 100% crops on the 7D due to the denser photosite placement - which suggests that you are both looking "more closely" than you would be with a 50D 100% crop and that the closer spacing might make "spill" of light more of an issue.</p>

<p>As I wrote... speculation.</p>

<p>Dan</p>

Link to comment
Share on other sites

<p>G Dan,</p>

<p>This shot, taken just for this thread, doesn't exhibit the kind of magenta fringing that Douwe is worried about so it is not really accurate to say <em>"I can tell you that the problem occurs on all Canon cropped sensor and full frame DSLRs that I have used in extreme exposure situations like the one you have shown"</em> as this one clearly does not.</p>

<p>It is a seascape very similar in lighting to Douwe's original and I deliberately overexposed by one stop to exaggerate the blown water. Full image to left, 100% crop of water highlights to right, obviously all untouched in post other than crop and convert to jpeg as the original was shot in RAW.</p><div>00V6XZ-194671584.jpg.8e443d462054266232ce04cc4f6d7984.jpg</div>

Link to comment
Share on other sites

<p> And so for me the 5D markii is back in the picture..<br>

I have found that the "purple fringing" phenomenon can also be generated with shots taken with a 5D2 - however there seems to be more to this than meets the eye. I remembered a couple of shots I had taken that showed this effect, your post prompted me to go back and investigate further. In the first attached Bridge screen shot you can clearly see two purple tinted shots and two normal looking shots. The purple ones were not purple until I reduced the exposure 1 stop in ACR! Now, if I re-open up the selected purple shot in CS4 raw converter it still looks purple, but just zoom in to 100% and the purple fringing almost disappears, screen shot 2. Convert the purple looking shot to jpg without making any adjustments and you get a shot devoid of purple fringing. Go figure.<br>

I've no explanation for what is going on here, but it seems the problem can be both generated and solved in PP! Scott, try reducing the exposure in ACR of your over exposed shot, my guess is that you will then see the the "purple fringing". I wonder if this is just a Canon sensor phenomenon?</p>

Link to comment
Share on other sites

<p> And so for me the 5D markii is back in the picture..<br>

I have found that the "purple fringing" phenomenon can also be generated with shots taken with a 5D2 - however there seems to be more to this than meets the eye. I remembered a couple of shots I had taken that showed this effect, your post prompted me to go back and investigate further. In the first attached Bridge screen shot you can clearly see two purple tinted shots and two normal looking shots. The purple ones were not purple until I reduced the exposure 1 stop in ACR! Now, if I re-open up the selected purple shot in CS4 raw converter it still looks purple, but just zoom in to 100% and the purple fringing almost disappears, screen shot 2. Convert the purple looking shot to jpg without making any adjustments and you get a shot almost devoid of purple fringing. Go figure.<br>

I've no explanation for what is going on here, but it seems the problem can be both generated and solved in PP! Scott, try reducing the exposure in ACR of your over exposed shot, my guess is that you will then see the the "purple fringing". I wonder if this is just a Canon sensor phenomenon?</p><div>00V6eo-194735684.jpg.3cb2dd48cc14df2a3b496c6498799398.jpg</div>

Link to comment
Share on other sites

<p>Scott, interesting example, and it certainly doesn't exhibit the fringing/blooming nor does it exhibit any CA. I'm impressed.</p>

<p>One thing to keep in mind about many of my posts - I often (but not always) choose my words somewhat carefully to be a bit less than absolute. For example:</p>

<blockquote>

<p><em>"I can tell you that the problem occurs on all Canon cropped sensor and full frame DSLRs that I have used in extreme exposure situations like the one you have shown"</em></p>

</blockquote>

<p>I didn't write that, for example, "the problem [always] occurs." And your shot is apparently proof of that and an illustration of my wording was correct.</p>

<p>The "problem" (though I think of that word as suggesting "an issue to deal with" rather than "a way Canon falls short") certainly is something that I find myself having to deal with on DSLRs ranging from an 8MP cropped sensor body to a current 21MP FF body. There are two ways (perhaps three?) that the "problem" can occur:</p>

<ul>

<li>CA "adding color" to scenes containing essentially randomized, highly detailed subjects with many edges between pure white and darker tones - like sparkling, sunlit water. This can often be improved in post by defringing and/or using the CA (red/green and yellow/blue) adjustments.</li>

<li>Purple fringing ("sensor blooming?") adding purple edges to the dark portions of boundary areas between very bright (typically blown out) and darker areas. The classic is branches against the sky. Fixing this in post is more complicated. Sometimes exposure decisions can help here.</li>

</ul>

<p>Several things can make this (the reflections on water) situation especially complex. Lens quality can come into play to the extent that CA adds color to otherwise nearly colorless areas. And a fringing can do this even with lenses that have low levels of CA.</p>

<p>I agree that it is possible to come up with shots that don't exhibit this. After seeing your example I went back through some of my older RAW files that include very similar subjects - brightly backlit reflections on water. I was going to do this to show examples of the "problem" on several bodies I've used. And I did find them. But I also found some examples that don't exhibit the problem and others in which it was completely resolved by some simple post processing steps.</p>

<p>Dan</p>

Link to comment
Share on other sites

<p>If the fringing is seen more on the higher photosite density sensor, there are a couple things to consider:</p>

<ol>

<li>It may seem more pronounced when you view the image at 100% magnification because you are effectively looking "more closely" at the image. A 100 x 100 pixel 100% crop from a 18MP sensor comprises a much smaller portion of the overall image than a 100 x 100 pixel 100% crop from a 8MP sensor. (In film terms, this would be sort of like inspecting one sample with a 10x loupe and then looking at the next sample with a 15 x loupe.) </li>

<li>Related to this, the flaw that is more obvious from the 18MP sample at 100% than from the 8MP sample at the same magnification may well affect a print image of a given size the same amount. A one pixel wide "flaw" in the 18MP image will be smaller in a print than a 1 pixel wide flaw from the 8MP image if printed at the same size.</li>

</ol>

<p>It is tricky to do direct comparisons between different photosite density samples (and between different format samples) at 100% on the screen. While it seems like looking at the samples at 100% would reveal the most "truth" about the images, it is also very misleading in several ways.</p>

<p>Now, none of this is meant to prove that there isn't a problem - I can't really do the objective comparison myself, and such a careful comparison might reveal an actual issue. But it is a caution about assuming too much from a sample.</p>

<p>Dan</p>

<p>(My idea of the best - but still not perfect - test would be to take two samples through similar ideal workflows to prints at the same size and then compare.)</p>

Link to comment
Share on other sites

<p>I would have thought that by now someone would have taken their 7D out and tried to replicate the OP's situation. We're getting into silly-land with answers. I'll try to get out some afternoon this week to see if I can replicate the situation with my 7D.</p>

<p>Dave</p>

Link to comment
Share on other sites

<p>Rick,</p>

<p>You are 100% correct. I'll try to post some examples later. If you take my shot and open in ACR and then reduce exposure by 1 stop it has a magenta fringe on screen at screen size (fit to screen) if you zoom in to 100% the fringe goes. If you print the file there is no fringing any which way.</p>

<p>This seems very strange to me.</p>

<p>I would like to know how many of Douwe's images have issues prior to development.</p>

<p>It seems that some images, mine and Rick's at least, can be made to exhibit magenta fringing even when the native capture does not, it is a post processing issue not a Canon issue, and I can't make the issue actually affect a print. My images are not shot with a particularly high density pixel sensor and on a full frame. It does not appear to be density or sensor size related but a pp screen rendering issue.</p>

<p>I have not had the time to do the same thing in DPP only ACR. I have several shots in the sequence above of different exposures, I'll do some more post on them and see what happens.</p>

Link to comment
Share on other sites

  • 4 months later...
<p>This is really sad. I am a professional photographer and shoot with my Cannon 50d a ton! I love that camera! However I have lots of problems shooting on bright sunny days with pictures showing lots of purple fringe. I came onto this site to find answers and what I hear is a ton of accusations. It could be the lenses, I have sigma and promaster lenses for the 50d just because Im in the woods with that camera a ton and dont want my exspensive L lenses damaged. I havent taken my 7d or 1d mark 3 outdoors just because Im rough on my outdoor cameras. But Im going to take the 7d out soon so I can compare the purple fringe phenom. For those who thinks this guy is bashing Cannon, he didnt say hey go buy Nikon or Sony he said he traded down to get better photo's. For those who think he is making it up there are lots and lots of people asking about the same problem. I know with my studio cameras I can adjust all of the lighting in the camera so that I dont have o worry about it. But Im inside I dont in there anyway. I would imagine that there is a way to do it with the 50d also, I just dont know how yet. Any how I will shoot with both soon and send in the results. I know this whole thread was done months ago but Im sure people still want to know.</p>
Link to comment
Share on other sites

  • 2 weeks later...

<p>The Magenta problem with the 50D is real. I have two 50D's, two 40D's, and a 1D Mark III. The Magenta problem is only visible with the 50 D. I tried the Canon Firmware update, and it's still there. I've tried combinations of Picture Styles, with custom settings, and it's still there. I've tried WB Shifts, and it's still there. I tried Arthur Morris suggestion regarding settings, and it's still there. I don't want advice regarding how to fix this in Photoshop. I want to fix it "in camera". I photograph wildlife, and use Canon EF L telephoto lenses....they are not the problem. Any advice including in camera adjustments that have worked for any of the members would be appreciated. I am not making this up, and have been fighting this problem for a long time. I just want to be as satisfied with my 50D's, as I am my other Canons. Thank you. Chris Webb (I listed my original Canon FD to EOS Converter Adapter for sale today...please look if you're interested). </p>

<p> </p>

Link to comment
Share on other sites

<p>Chris,</p>

<p>Did you do any exposure adjustments to your images prior to seeing the magenta issues? Can you post an example. If you read my earlier answer then you can see that my 1D files can be made to do it, but only on screen after exposure adjustments. Do they print magenta?</p>

Link to comment
Share on other sites

  • 1 year later...
<p>I got some really bad purple fringe taking a backlit shot with my 7D and 50mm lens (ISO400, f 1.6, 1/80sec)....I am working in Lightrooom, definging made not difference, decreasing the purple saturation and increasing highlight toms and light tones were the only way to get rid of it....although I lost the edge definition.....</p><div>00ZGZL-394551584.thumb.jpg.9cff455c540749df706ccfca0930f0e6.jpg</div>
Link to comment
Share on other sites

<p>Yes, I zoomed in (in Lightroom), and it's there!...this is pre-processing, the image was taken in RAW and I just exported it as JPEG in Lightroom....I will try in another program?<br>

I have more photos where I used a little off camera flash with the same settings and the purple is completely gone (the blinds are much more visible), also some where I increased to ISO500 and 1/125 and although there is a little purple, the effect is not nearly as bad....</p>

Link to comment
Share on other sites

<p>Karen,</p>

<p>You have a completely different issue, the light through the window is way way too bright, the dynamic range is vastly outside the capabilities of the sensor. You need to lower the window light or raise the light inside to better balance the interior and exterior brightness. Basically if the highlights are over four stops over your exposure you are going to get problems like this.</p>

<p>Probably close to impossible to correct well in post, but not too difficult to prevent at time of capture.</p>

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...