Color Fringes, Banding...how to solve the problem?

[laurentlacoste]

<p>I've just bought a new camera and went to try it yeasterday on the beach. On a few of the pictures, strong and large color fringing appeared on the raw files. What could be the cause of it? How can I solve the problem?<br>

Here's an attachment of a cropped picture as an example.</p><div></div>

[benbangerter]

<p>The image you posted looks fine to me. What specifically is your complaint?</p>

[vasilis1]

<p>It looks fine to me too. Could it be that you cropped out the faulty part of the picture?</p>

[guido_h]

<p>If you mean the slight red/cyan hue differences in the left upper area of the sky, which is overexposed - that's simply because with digital sensors, the three (R/G/B) channels blow out differently, resulting in color shifts in highlights as one channel after the other clips. You can verify this with the Levels dialog in Photoshop.</p>

<p>The solution is to avoid overexposed areas in your image. In post-processing, you can reduce the banding by adding a small amount of noise artificially (Filter / Noise / Add Noise command in Photoshop).</p>

[laurentlacoste]

<p>Ben, Vasilis, It is clearly visible on my monitor (24 Imac, calibrated with the Eye One Display). I'm surprised that neither of you sees the effect? I tried to print the picture, it is also visible there.</p>

[laurentlacoste]

<blockquote>

<p><br /> If you mean the slight red/cyan hue differences in the left upper area of the sky, which is overexposed - that's simply because with digital sensors, the three (R/G/B) channels blow out differently, resulting in color shifts in highlights as one channel after the other clips. You can verify this with the Levels dialog in Photoshop.</p>

</blockquote>

<p>Guido, thanks for the explanation. I'd never ahd that kind of problem so far with the D2X. Does that mean that the D5000 handles the problem in a less efficient way? Judging from my settings the sky wasn't clipped, except for the extreme left-hand side of the picture. Thanks for the PS tip. I'll try that.</p>

[guido_h]

<p>Laurent, I only have experience with the D300 (which is closer to the D2X than to the D5000), so I don't know what goes on inside Nikon's more recent models nowadays. I would guess that the D5000 should perform at least as well, for the sake of progress :-)</p>

<p>You might want to tweak the Camera Raw settings which you used for importing the raw file. Maybe some of the clipping was introduced there? I suggest to play with the "Recover" and "Exposure" sliders to find out exactly how much raw information was actually captured in the clipped sky areas.</p>

<p> </p>

[roger_smith4]

<p>I'm not really seeing banding on my calibrated CRT. I'd guess it's your monitor.</p>

<p>I also agree that the file looks overexposed and overdoing the "recovery" setting can introduce ugly edges between the normally exposed and overexposed areas of your film.</p>

[laurentlacoste]

<blockquote>

<p>I'm not really seeing banding on my calibrated CRT. I'd guess it's your monitor.</p>

 

</blockquote>

<p>Roger, I don't know because as I said the ugly edges and yellow and red bands appear on my two monitors and also on the prints. Plus, I'd never had that kind of problem before with my previous camera (D2X). Although discreet, the defect is still here and visible when one pays attention. But then Guido might be right when he suggests that the way the sensor responded there is the cause. This was a strongly backlit sunset scene, hence the overexposed sky (not clipped though). I must say this is quite disappointing, all the more since the D5000 seems to be a really good camera. I'd like to find out what happened then.</p>

[phule]

<p>[[Roger, I don't know because as I said the ugly edges and yellow and red bands appear on my two monitors]]</p>

<p>If other people are not able to see it, perhaps you should do a screen capture of the image, and upload that, instead of uploading the JPG itself.</p>

<p> </p>

[guido_h]

<p>It was quite easily visible on my monitor. If you check the bands of clipped pixels in PS "Levels", and then look at the image again, maybe it's more obvious then.</p>

[john_demott]

<p>Laurent,<br>

I believe that the D5000 uses a compressed NEF format while the D2X uses uncompressed NEF. Roughly, that is the difference between a 10 bit file and a 12 bit file, with most of the "lost" information being in the highlights. I don't think you should ordinarily see a difference unless you have used some fairly heavy post processing adjustments to recover highlights. If you have an area of highlights that is nearly blown out, and you use post processing to recover the area, then you are "stretching" the curve by enhancing the difference in values of pixels that were recorded as very close in value...the result can be banding. </p>

<p>Questions: What raw converter did you use and did you use any sort of highlight recovery? The highlight recovery in Adobe Camera Raw can be very effective but since it depends on recovering information from other color channels, it can lead to some unfortunate color effects. </p>

<p>A suggestion for fixing your particular photo would be to use the gradient tool in Photoshop to create a gradient from the lightest color to the darkest color in the sky and overlay the gradient on your image, using the Blend If sliders so that it is only visible in the area of the sky. That should give you smoother transitions without visible banding.</p>

[laurentlacoste]

<blockquote>

<p>If other people are not able to see it, perhaps you should do a screen capture of the image, and upload that, instead of uploading the JPG itself.</p>

</blockquote>

<p>Rob, here's a screen capture. I tilted back the monitor on my laptop to exaggerate the effect for you to see. As I said, it is quite discreet and one has to pay attention to notice it, but it's here nonetheless.</p>

<blockquote>

<p>Questions: What raw converter did you use and did you use any sort of highlight recovery? The highlight recovery in Adobe Camera Raw can be very effective but since it depends on recovering information from other color channels, it can lead to some unfortunate color effects.</p>

 

</blockquote>

<p>John, thanks for the info and the tips, very much appreciated. I used NX2 to convert the image to 16-bit TIFF before opening it in PS for editing. I just did small curve adjustments before noticing the band effects. No highlight recovery done.</p><div></div>

[roger_smith4]

<p>On my low-quality (but hardware calibrated) Viewsonic LCD I see slight banding. On my high quality and hardware calibrated CRT I didn't see banding.<br>

<br /> To better troubleshoot we've got to figure out what's going on at the printing stage and why there is visible banding there. Can you share what printer you are using and the settings you are using with it? Are you using a ICC printer profile for the ink and paper you are using? Can you soft proof in photoshop and look at different rendering intents (perceptual, relative colormetric and saturation) to try to find one which bands the least? Are any parts of the image out of gamut?<br>

My guess is that to get a decent print you'll want to make sure you are using a good printer profile and appropriate driver settings, and might want to take a good section of the sky and use it to make the sky more uniform, getting rid of the transition that is causing banding. You might also try redoing your edits in NX or taking the image into Adobe Camera Raw to see if a different treatment of the raw file where you expressly try to avoid banding yields a better result.</p>

[john_demott]

<p>Laurent,</p>

<p>I don't see any obvious reasons for the banding. Did your raw conversion include any steps to reduce brightness, either overall or in the area of the sky? To me, the most likely culprit for this kind of problem is some sort of post processing adjustment that amplifies otherwise minor differences in tonality and color across the area of the sky. Were the bands visible in NX2 before you converted to a Photoshop file? Were all the files that exhibited the problem basically the same...i.e., banding in the area of the sky with an exposure close to blowing out? </p>

<p>Have you tried to replicate the problem in a systematic way by taking a series of bracketed exposures including an area of sky, perhaps with various atmospheric conditions (I'm wondering if the hazy conditions could have played a role)? It might be interesting to try replicating the problem while shooting an identical series with your D2X...it would give you a clue whether this is a problem unique to your D5000.</p>

[laurentlacoste]

<blockquote>

<p>To better troubleshoot we've got to figure out what's going on at the printing stage and why there is visible banding there.</p>

</blockquote>

<p>Roger, thanks. I have an Epson 1800 printer using Ultrachrome inks. My workflow from monitor to printer is calibrated and I use ICC profile for the papers I use. I've never had any problem with the printer so far. Basically, what I saw on the screen was reproduced on the prints.</p>

<blockquote>To me, the most likely culprit for this kind of problem is some sort of post processing adjustment that amplifies otherwise minor differences in tonality and color across the area of the sky.

<p>(I'm wondering if the hazy conditions could have played a role)?</p>

 

</blockquote>

<p>John, I tend to think that the explanation lies there. I must have done something wrong somewhere. I'll work on the images checking again all the raw files first to see if either I've pushed levels too far or if atmospheric conditions might have played a role during shooting, but I don't think I will have the opportunity to revisit the place soon. it is true that the lighting conditions were quite special, very intense sunlit end of day seashore light with a hazy white layer. Thanks again for helping me out here.</p>