Jump to content

Strange Digital Banding in Skies - Is that normal?


vanessa_b2

Recommended Posts

<p>I hope I've posted this query in the right category as I'm not sure if it's a Nikon thing, a Nikon lens thing, a digital thing or something else - something I'm doing.<br>

I use a Nikon D7000 and the lens I used for the picture attached was the standard Nikkor 35mm lens 1:1.8G. I took this picture on the highest resolution possible as a jpeg.<br>

If you look at the sky, you can see a sort of banding of the different shades (you might need to tilt your screen). This often happens with pictures containing sky. I really do not like this at all and I'm wondering if this is the norm for this camera with a good lens like its standard lens. I had pictures printed onto postcards some time ago at high definition and this actually showed. It isn't obvious, but to me it just doesn't seem a high quality picture.<br>

Is this what I have to expect from this camera or lens? I can't afford a better camera/kit right now. Or maybe it's something I haven't got right? Please tell me what you think it could be. What's strange is the results are inconsistent - sometimes it happens, sometimes it doesn't and I can't figure out why. It's not because of the direction of the lens in relation to the light though I've noticed it can be worse when facing the light source. This picture wasn't.<img src="http://www.photo.net/photodb/member-photos?user_id=5384673%20" alt="" /><br>

Thanks</p><div>00a9Ee-450647584.jpg.001c6bb04500f80e8e40aa4907f1d352.jpg</div>

Link to comment
Share on other sites

<p>Unfortunately, the way you created the link to your photo wasn't right, so we can't see the image.<br /><br />But most of the time, what you're describing is a result of JPG compression - certainly not a function of the lens. Are you shooting right to JPGs straight out of the camera, or are you creating the final JPGs through whichever software you're using, after the fact?<br /><br />Not being able to see the photo, it's hard guess whether or not you're also talking about lens flare or some other actual optical artifact. Can you try attaching a copy of the file to another reply to your post?</p>
Link to comment
Share on other sites

<p>The blue channel in the histogram looks blown in the highlights. A blown channel means false colours in the affected region, and integer changes of the value near the point of blowing in areas with smooth transitions such as your sky often result in such banding. So try to reduce exposure. If you shoot NEF the highlights might be recoverable. <br>

Notice this is more obvious on poor quality low-bitdepth displays. Much more obvious on my laptop display than on the external IPS one. Display calibration tends to make it worse on poor displays. <br>

There is also some jaggedness in the histogram, possibly due to processing (ADL?) and JPG compression.</p>

Link to comment
Share on other sites

<p>Thanks for your responses so far. I've just tried attaching the link to my gallery again, because that image is higher res than the one I had to attach separately so you'd see it better. I'm following the instructions here on Photo Net but it's just not having it.</p>

<p>The original was the highest-res jpeg. I didn't change the jpeg from my camera other than reduce it for this forum - but the large original shows the same. I know RAW is better in most cases, but I'm not convinced this should happen normally with a highest-res jpeg. Will think hard about the points you've raised.</p>

<p><img src="../photodb/member-photos?user_id=5384673" alt="" /></p>

<p> </p>

Link to comment
Share on other sites

<p>The problem you describe is called "banding" and is closely related to "posterization". Primarily, it comes from working at 8 bits per channel instead of 16 bpc, especially when one makes large changes to an 8 bpc image in postprocessing. Unfortunately, it is fairly common. However, because it is common, there is a lot of material available on the web that discusses this problem. For example, see:</p>

<p>http://www.cambridgeincolour.com/tutorials/posterization.htm<br /> - - - Good intro to the problem.</p>

<p>http://www.photo.net/digital-darkroom-forum/00ZoFv<br /> - - - See my post of Jan 2nd in this thread for more info.</p>

<p>http://www.photo.net/digital-darkroom-forum/00Vl7k<br /> - - - Even more detail, incl. a discussion of how to try to minimize the problem after the fact.</p>

<p>HTH,</p>

<p>Tom M</p>

Link to comment
Share on other sites

<p><<I know RAW is better in most cases, but I'm not convinced this should happen normally with a highest-res jpeg. >></p>

<p>it has nothing to do w/ 'highest res jpg'. In fact, going 'hi-res' (read: hi quality setting) probably makes it worse since it puts extra emphasis on high frequency edges. In 99% of cases, you dont need 'highest quality' jpg rendering. It makes the file much larger w/ essentially zero visual improvement.</p>

Link to comment
Share on other sites

<p>Vanessa, shoot in RAW and convert to 16 bit TIFFs for the best archival quality with no banding. The fact that some LCD monitors only work internally with 6 bits/channel precision can exaggerate the problem. Have you tried viewing the images on a higher quality display?<br>

<br /> I'll take this opportunity to have another little rant about JPEGs. The JPEG file format was NEVER meant to be used for high quality image archiving. It was designed for efficiently transferring compressed and compact medium quality 16.7 million colour images over the internet. Why it's become the de-facto standard for storing supposedly high-quality digital camera images is beyond comprehension.<br>

<br /> Incidentally Tom, banding and posterisation are one and the same thing. Posterisation is defined as a splitting of a smooth and continuous tone or gradation into discrete areas of solid tone or colour - i.e. banding. Different words, same meaning. The only difference being that posterisation is sometimes done deliberately for graphic effect, while banding isn't.</p>

Link to comment
Share on other sites

<p>Rodeo, it's a small semantic matter, but I've got to slightly disagree with you. </p>

<p>Clearly, both arise from the same cause and we completely agree on this. </p>

<p>The term, "posterization", can be used to describe this effect even in highly complex / textured images. In contrast, to see actual banding, (ie, big, obvious bands of color that run across substantial portions of an image), and hence, to use that term correctly, one needs to have, as you point out, areas of smooth and continuous tone or color gradations, most commonly, the sky. </p>

<p>The distinction is not a big deal, but because no one would ever call posterization in a complex image "banding", (because there are no big bands), I prefer to keep the uses of the two words separate. If I remember correctly, I think the Wikipedia and other articles on the subject preserve the distinction in technical discussions, letting the distinction slip in less technical discussions.</p>

<p><--- Minute examination of the back of our semantic fingernails has concluded. We now return to your regularly scheduled programming. :-) ---></p>

<p> </p>

Link to comment
Share on other sites

<p>My camera only has 12 or 14-bit . My version of Photoshop (please don't ask) won't allow too much editing in 16-bit. When I open a Tiff in photoshop, it automatically goes into 8-bit.</p>

<p>If I record the picture in Adobe RGB colourspace, and later edit in it in a program that will only let me work on it as 8-bit, does that mean I damage the picture because Adobe colourspace requires 16-bit?</p>

<p>I know RAW is best for archival reasons, but I'm not printing at enormous dimensions. We're looking at a massive difference in quality over a problem that doesn't need to happen. Then again if I record in a lower quality jpeg, I won't be able to print at a certain size. Sometimes it's difficult to know how deep to wade into all the technicalities without losing the ability to sort what's practicable and what isn't.</p>

<p>I've been reading through the articles and there's a lot to take in, although it makes sense.</p>

Link to comment
Share on other sites

<blockquote>

<p>I know RAW is best for archival reasons</p>

</blockquote>

<p>That's not really it. It's important because there is <em>far</em> more data there, than in a JPG. Each pixel has much more latitude for you to make adjustments. You don't want to go 8-bit until you're done with those adjustments.</p>

<blockquote>

<p>Then again if I record in a lower quality jpeg, I won't be able to print at a certain size.</p>

</blockquote>

<p>The JPG isn't going to have any fewer pixels than the RAW file. A properly exposed image handled according to JPG processing rules that you've intelligently set up in the camera to suit the circumstances in which you're shooting and with a specific output use in mind is going to print just as nice and just as large as a post-processed RAW file. The problem is that it can be hard to get everything optimally perfect for that limited-depth JPG on the fly. So the answer is ... don't. Or, shoot RAW + JPG, and get the best of both worlds.<br /><br />Download a free trial copy of Capture NX2, and work with some RAW output from your D7000. You will be very surprised.</p>

Link to comment
Share on other sites

<p>Vanessa, there are two main aspects to JPEG files; their size in pixels (pixels high x pixels wide), and their compression level or "quality". The size in pixels doesn't affect the colour quality, but the compression level does. It's important to get these two things separated in your head. In fact I would completely forget about the fact that you can set your camera to shoot smaller images than the maximum 16 Megapixels. Because why would you need to? You can always resize your images on the computer afterwards.</p>

<p>The quality setting of JPEGs is much more important, and unless card space is an issue, again there should be no need to shoot at anything less than the highest quality (fine). Even so, one way that a JPEG file saves memory space is to throw away some colour information. It does this partly by dividing the image into little blocks of pixels, each with a "local" pallette using a restricted number of colours - a small subset of the 16.7 million possible 8 bit colours. Anyway, without getting too technical: JPEG bad - RAW good. RAW holds all of the information captured by your camera and JPEG doesn't. Your camera may "only" capture 14 bit images, but that's 2 million times more colour subtlety than a JPEG could possibly contain.</p>

<p>Re AdobeRGB: The Adobe colourspace doesn't require 16 bits per channel, but most display devices won't handle the Adobe RGB gamut, therefore some conversion of saturation levels has to occur when showing AdobeRGB images. For example; a monitor profile will automatically convert another colourspace to sRGB or something similar. This conversion particularly affects the Cyan area of the spectrum corresponding to some sky tones. A colour space conversion always throws some image information away, because all image manipulation is damaging - tone or colour is never created, only destroyed.</p>

<p>Speaking of colour profiles. I suspect you have a chopped down or early version of PhotoShop. Some early Windows versions didn't handle colour profiles at all well, especially if you've used the awful Adobe Gamma monitor "calibrator". Far better to calibrate your monitor in hardware and disable Adobe Gamma, which causes noticeable posterisation in shadow tones. In fact many profile combinations will cause problems in some versions of PhotoShop and I second the advice to try NX2.</p>

Link to comment
Share on other sites

<p>I think it's going off at a tangent perhaps because I replied to points made by different people in the same posts.</p>

<p>I do understand what RAW is as oppose to jpeg. And I should probably always shoot in RAW (not just sometimes). When shooting in jpeg I've chosen quality over size so I'm not sure where that discussion point comes from. Maybe I haven't got a grip on the lexicom. If I always shot using RAW then I would decrease the chances of this banding (which I now will :-) ).</p>

<p>My point was I wanted to understand why this was happening at all, and why sometimes and sometimes not. I wanted to be able to understand why this happens when it doesn't necessarily have to, whether or not I then decide always to use RAW. In one of the articles Tom linked it states "working in colour spaces with broad gamuts can increase the likelihood of posterization because they require more bit depth to produce the same colour gradient" which was why I was wondering if there was any point working on large 16-bit TIFs if that happens. I am pretty sure my monitor/calibration is fine and has little to do with it (btw its default is sRGB).</p>

<p>I understand about 8-bit jpegs and RAW and working on best file possible before "downgrading" for its ultimate destination but if I am in a position where there isn't anything I can do because I can't upgrade all the other hardware/software, it would be a shame not to work with perfectly usable jpegs from which I've printed successfully before rather than do nothing at all because I'm not able to edit RAW files.</p>

<p>Having said that I will certainly look into Nikon Capture, otherwise I have to work with what I've got. You've all given me plenty to go with here, thanks guys.</p>

Link to comment
Share on other sites

<blockquote>

<p>"working in colour spaces with broad gamuts can increase the likelihood of posterization because they require more bit depth to produce the same colour gradient"</p>

</blockquote>

<p>OK. Let's try and clear this up once and for all. I think what the above statement refers to is using a wide gamut colour space with only an 8 bit/channel colour-depth. That's like trying to squeeze a quart into a pint pot, and isn't a good idea IMHO. Sixteen bits per channel (48 bit colour) is a different matter. You then have 65536 levels of tone per channel, or 65536^3 different colours to play with. Consequently you can do a lot more "throwing away" of levels before the result becomes visibly degraded.<br>

Even 14 bits from the camera gives you 16384 different levels of tone per channel, but these have to be saved in a 16 bit/channel file format, since there's nothing between 8 and 16 bits to choose from. In short, you can practically decimate the 14 bit RAW files from your camera before you even approach the limited range of tones a file format like JPEG can hold.</p>

<p>I know that a lot of options have been thrown into the mix. E.g. Profiles, monitor hardware, colour spaces, processing software and file formats. However these things are all cumulative, and you need to methodically eliminate the effect of each of these. So no quick and easy answer I'm afraid.</p>

Link to comment
Share on other sites

<p>Hi, Vanessa.<br>

I had the same problem with a Nikon-refurbished D700. Nikon's west coast service facility in El Segundo, CA eliminated the bands under terms of the 90-day warranty, suggesting that banding is a problem that Nikon knows how to eliminate. </p>

 

Link to comment
Share on other sites

<p>I would think Conrad is right. I have never seen this banding in my jpegs from D300. I shoot "raw + jpeg fine" as standard. Here is some of my camera settings:</p>

<p><strong>In the shooting meny:</strong><br>

JPEG compression: Optimal quality<br /> NEF recording: Off (no compression of the files)<br /> NEF bit dept 14-bit (highest possible in D300)<br /> Picture control: Standard (not vivid, highlights get easier blown out)<br /> Color space: sRGB<br /> <br /> I use ViewNX2 as the raw converter (it's free). The .nef raw files are by default converted to look exactly like your jpeg's straight from the camera. I use ViewNX2 to convert the .nef files to .tiff files, which I then open in Gimp (it's free) and process further there. I save the files in .xcf format in Gimp. Can also be saved as .tiff files, but you will loose any layers then. You can also convert the .jpeg files to .tiff files in ViewNX2. And that might be a good way to do it to be able to process the .jpeg's with as little quality loss as possible. When you're done, you save the files as .xcf files and .jpeg files.</p>

Link to comment
Share on other sites

<p>Hi Vanessa -</p>

<p>a) The image that you posted has gone through several processing steps, each of which can impact the artifacts we see in the sky. To try to separate out the various possible causes of artifacts, and be able to unambiguously answer your question, please do the following: Open the original JPG that was produced by your camera in your editor. Make absolutely sure you are opening the original file that came out of your camera, and not a version you saved using your image editing software. DO NOT modify the image in ANY way whatsoever, except to crop out a portion of the sky that is about 500 pixels in both directions. Resave the cropped version as another JPG using the highest quality setting that keeps the file size under 300 kB. Please post that image in this thread, and tell us the JPG quality setting that you used to keep the file under 300 kB in size.</p>

<p>b) Take a look at the attached image. It is the sky portion of the image that you used to start this thread, except that I processed it to greatly exaggerate the artifacts. Note the red square. It is exactly 8 of your original pixels in both dimensions. Note that almost all of the artifacts that you see are either exactly this size (ie, 8x8 pixels) or some multiple of this size, eg, 32 pixels horizontally by 8 pixels in the vertical direction, 4 px by 4 px, etc. etc. The reason that rectangles of these sizes are so common in your image is that they were caused by using a low JPG quality setting at some point between your camera and us seeing the image. Your image editor may have introduced these, your camera may have introduced these, or the photo.net uploader may have introduced them. By exactly following the procedure I outlined above, the photo.net uploader will not modify your image, and the process of re-saving the cropped version using your image editor should have hardly produced any artifacts. Simply going through these steps will answer a lot of questions.</p>

<p>c) Blocky artifacts like these have absolutely nothing to do with banding, only with JPG compression. Once you post the 1:1 cropped image described above, we'll be in a better position to discuss the other artifacts present, ie, (i) the horizontal banding and (ii) the portion of the image that I outlined in yellow.</p>

<p>Regards,</p>

<p>Tom M</p>

Link to comment
Share on other sites

<p>Vanessa, maybe you could check out what happens when you change the jpeg compression in your camera.<br>

<br /> Go to: Menu - Shooting menu - JPEG compression. Choose the option "Optimal Quality". Click "OK" and take some new pictures. Does the problem still persist?</p>

Link to comment
Share on other sites

<p>Banding isn't just a digital issue, I've experienced it in the past on a few occasions with my Fuji GSW690II on Velvia; perfectly exposed transparencies with very visible bands of different shades of blue in the sky. Maybe to do with the angle of the sun in the sky relative to the camera or polarising filters?</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...