Nikon D600 horizontal banding at high ISOs?

<p>I recently got a D600 for Christmas (what an awesome present!).</p>

<p>When shooting at ISO 6400 and up, there is horizontal banding visible in the shadows (around a stop or so below neutral). It is not terribly visible at 6400, but is easily visible at 12800, both in Lightroom and in the in-camera preview.</p>

<p>My friend who got me the camera also got one for himself, and it shows exactly the same behavior. The serial numbers differ by only a few hundred, so they may be out of the same batch.</p>

<p>My question is this: is this normal for these cameras? I'm coming from a Canon 7D, which shows no such artifacts at all even at ISO 12800 (and that's a crop camera with earlier sensor tech, albeit from a different manufacturer altogether).</p>

<p>I can post example 100% crops if necessary or useful.</p>

<p>Thanks in advance for any responses.</p>

 

<p>On the D600, the top rated ISO is 6400. Hi 1 (12800) falls into the "better than nothing" category. I wouldn't expect much once you are above the rated range.</p>

<p>Wait. This isn't normal. Horizontal banding is very unusual on the Exmor cameras. They use a column readout, so the /very/ little banding there is, is usually vertical. <br>

I just shot an album cover at ISO25,600 equiv on the D800. The results, including colors, were very good, good enough for production work. The only issue I've run into with the D800 is thermal noise. When working at those settings, I do a black frame subtraction to mitigate the thermal noise. <br>

Do you have a sample of the banding that you're describing that we could examine?</p>

<p>ISO 12800:</p>

<p><a title="DSC_0119-cropped by n2185x, on Flickr" href=" src="http://farm9.staticflickr.com/8231/8356392970_c430850bf9_b.jpg" alt="DSC_0119-cropped" width="1024" height="683" /></a></p>

<p>ISO 6400:</p>

<p><a title="DSC_0098-cropped by n2185x, on Flickr" href=" src="http://farm9.staticflickr.com/8045/8356393066_375ef9c68b_b.jpg" alt="DSC_0098-cropped" width="1024" height="683" /></a></p>

<p>The above are 1024x683 images, and show up smaller here, so the banding isn't quite as obvious as it is when you're viewing at 100%. The two images link to the originals on Flickr, so you can go there and view the different sizes. Just view the largest size and you'll see the banding in all its glory.</p>

 

<p>Note that the above two images are 100% crops, not the full images.</p>

 

<p>Given the near-to-black values in these 100% crops, I think you're actually getting pretty good results. And indeed, I might have been mistaken about banding only being a vertical issue. I don't think my D4 does better than this on pattern noise. I don't think this is a service issue.</p>

<p>I had completely forgotten that I had a couple of ISO 12800 samples from the D600 in my archives. Noise level is certainly serious.</p><div></div>

<p>Crop</p><div></div>

<p>Noise isn't the problem for me. The banding is. I have no problem with random noise, because that is easily dealt with. Banding is more of an issue, and takes specialized software to deal with and a separate (and slow) postprocessing step.</p>

<p>My question is solely about the banding, not the noise itself.</p>

<p>Note that while the banding is more visible in the darker regions, there are hints of it in the lighter regions as well.</p>

<p>I'm not going to worry about it if it's normal, but I'm beginning to think that it's not. Shun, your shot shows noise but no banding that I can see.</p>

 

Kevin, since your image sample has so much dark area, I am afraid that you are expecting too much.

 

Generally speaking, I would try not to use the top rated ISO and never venture into the extended, hi range.

<p>The dark areas show the banding the most. But it is present in lighter areas as well (see the left side of the ISO 12800 shot, for instance).</p>

<p>The entire point of the thread is to determine whether what I'm seeing is normal for these cameras or not. The only way to know is to see ISO 12800 shots from other D600 cameras that cover the same tonal range. If those shots show no banding, then it means that my camera is <em>not</em> normal.</p>

<p>If my camera is normal, then I'll obviously just live with the issue. If it's not, then I'll either exchange it (if possible) or send it to Nikon for service. But I need to know one way or the other whether what I'm seeing is to be expected.</p>

<p> </p>

<p>I grabbed the ISO 12800 RAW sample file from dpreview.com's review of the D600. I'm seeing at least some of the same thing there as I'm seeing in mine, at least in some areas. So it looks like this may be normal for this camera (the 5D3's RAW is notable for its lack of banding in the same areas).</p>

<p>Note that I'm using Lightroom 4.3 for postprocessing. It may be that other raw converters eliminate the banding automatically. Shun, which raw converter are you using?</p>

 

<p>I have to say it, because some one might as well.</p>

<p>Are we really discussing a noise related issue, at ISO 12800 and at 100% crop ? I had to stare at the larger version of the linked image to see the banding, and that is probably only due to being TOLD to find it. Come on people. </p>

<p> </p>

<p>What John said. I'm sorry, but I <em>still</em> can't see the banding, after studying the crops for a minute or so.</p>

<p>(A couple of minutes later) . . . Are we talking about the very tiny bit lighter area right near the bottom of each crop? I can sort of see it if I scroll my screen up and down. If that's what you're looking at, is it visible in a print of the whole image?</p>

<p>What banding? I can't see any. Or maybe I have a lower expectation.</p>

<p>It's easier to see if you click on the images, then tell Flickr that you want to view the original size (1024x683).</p>

<p>It's not horrible by any stretch. However, I've gotten spoiled by my Canon 7D, which exhibits no banding whatsoever at high ISOs.</p>

<p>Horizontal banding at high ISOs used to be the norm. Canon's sensors stopped exhibiting that as of the 18 megapixel crop sensor that's in the 7D. Everything in their line later than that exhibits no high-ISO banding at all. Everything earlier than that does.</p>

<p>I thought, perhaps, that Nikon, too, had eliminated it from their cameras. Seems, perhaps, that they didn't after all. Such is life. There's no such thing as a perfect camera, and it seems there's no such thing as an artifact-free sensor, either. In trade for eliminating high ISO banding, Canon cameras now exhibit vertical banding in the deep shadows at low ISOs (even ISO 100), and that makes pushing the shadows even a bit something of a problem with those cameras. The Exmor sensor cameras, on the other hand, are glorious in this respect. You can push the shadows by 5 stops and the end result is <em>clean</em>.</p>

<p>If the choice is very high dynamic range at low ISOs + high ISO banding versus mediocre dynamic range at low ISOs + lack of high ISO banding, I'll take the former, because the latter can be cleaned up, but there is no substitute for dynamic range.</p>

<p>To be clear, the purpose of this thread is not to complain about the banding, but only to ascertain whether it is normal for the D600 to exhibit it at these ISOs. I was hoping that Nikon (perhaps through Sony) had licked the problem of sensor artifacts once and for all. Looks like they're not there yet. That's fine in this case, because the banding can be removed with an additional postprocessing step (Nik Software's DFine is fabulous for this), and it's only really present at ISO 12800 and above.</p>

 

<p>If you think that's bad banding Kevin, then you ain't seen real banding!</p>

<p>IMHO what's visible in those tiny crops is nothing but "artefacts of randomness". Any random texture will throw up areas where patterns can be seen in the noise. Like people claim to see human or elf-like faces among densely packed tree leaves, or a religious image on a piece of burnt toast. The "lines" don't even appear straight to me, but more gently curved and slightly angled to the horizontal. All of which mitigates against this being true banding.</p>

<p>In any case, it's such a low-level and localised effect that it's not worth bothering about, and certainly doesn't look like a sensor or camera fault to me.</p>

<p>A) I don't see any banding. But then my monitor is calibrated which means that it is much darker than typical. Which in turn means that, if it is really there, it won't show up in prints.<br /> B) The 'horrific' noise at ISO 12800 with this camera reminds me of ISO 400 Tri-X film - at least when viewed at 100% crop. This is truly amazing technology.</p>

<blockquote>

<p>The 'horrific' noise at ISO 12800 with this camera reminds me of ISO 400 Tri-X film - at least when viewed at 100% crop. This is truly amazing technology.</p>

</blockquote>

<p>Sure, but technology has improved drastically in the last few years. The last time I shot Tri X was probably over 30 years ago, as I don't shoot much B&W anyway. However, I even find DSLRs from over 5 years ago not very acceptable any more.</p>

<p>The is what the D600 can do at ISO 100 at that same scene. It was the same 50mm/f1.8 AF-S lens at f5.6, but the shutter speed was 1.6 seconds for ISO 100.</p><div></div>

<p>That room must have been VERY dark.</p>

<p>John, I tend to go to that location to test indoor, high-ISO results from various cameras. It is a local medical clinic and they open at 6:30am every morning. I typically go there before 7am on Saturday or Sunday so that there is usually nobody around and I can set up my tripod to make time exposures. It gets busy later in the day.</p>

<p>I would try to elminate some possibilities it may be radio frequency related (or some other type of frequency). What caused it in some models (other brands to etc.) was the focusing or vibration reduction motor would release a tiny amount of RF/interference. The sensor and the camera's electronics pick up this interference (the more metal the camera body & lens the less susceptible) and normally it's not enough to be detected but at high ISO everything is amplified and the interference gets amplified as well and at especially high ISO starts to show up as banding. This happens because the interference is a sine wave with peaks & valleys. The sensor reading line by line if reading happens at the peak of the interference wave then the line is artifically boosted. Reading at the base the interference is less so the line comes out darker. Rinse and repeat and banding is created. </p>

<p>I would try seeing what happens if you shut off as much as you can... vibration reduction, autofocus, no flash (make sure it's off not just sleeping), see if you get banding. Then turn them on and see if it's different. You may even want to try a different lens one type of lens its motor (I don't know if Nikon) released interference that would cause banding while a different line did not. </p>

<p>Not saying the above will help but it's worth a try. There's a lot of things that can cause it.</p>

<p>That's a good point about interference. I'll see what I get when turning everything off.</p>

<p>If it really is RF interference, then I would expect to see a lower frequency pattern to the banding (the chance that the interfering signal's frequency exactly matches that of the readout is extremely low). That's not something I've looked for yet. Regardless, it's a good thing to eliminate.</p>

<p>Nope, banding remains even with all the good stuff turned off.</p>

<p>Also discovered that it's visible in the dark areas at ISO 3200.</p>

<p>At what ISO should I <em>not</em> be able to see banding at all?</p>