Nikon D700 Banding Problem

Discussion in 'Nikon' started by michael_seika, May 26, 2009.

  1. I have had my D700 for about eight months now and I have not noticed this problem until last week when I was in Paris trying to take, for the first time, handheld night photos of the city. I was using ISO 6400 and I noticed serious banding on the photos, as shown in the attached typical example. Is this normal for this $3000 camera or is there a problem with the camera? Thank you.
    00TTdv-138137584.jpg
     
  2. Another example...
    00TTeA-138137684.jpg
     
  3. ...and another one. Thanks.
    00TTeC-138137784.jpg
     
  4. I wouldn't say that is normal. I would say it might need to take a trip to Nikon to get fixed. You might go ahead and download the latest version of firmware 1.01 and see if it helps, if you haven't already done it.
     
  5. The banding appears only in cases having high contrast like these ones.
     
  6. What on earth are you both talking about ? What banding ? I see no banding in any of the examples, only some variation of light from the spotlights.
     
  7. "What on earth are you both talking about ? What banding ? I see no banding in any of the examples, only some variation of light from the spotlights."
    I agree.
     
  8. I can see it. (I'm using a color-managed browser on a calibrated screen; that may help.) It's pretty subtle. The bands go vertically on the vertical shots, and horizontal on the horizontal shots. There are maybe 20 of them across the width of the first frame (I didn't actually count), approximately evenly spaced.
    I think they're not actually evenly spaced -- they're located where the blown highlights are located.
    I haven't noticed this from my D700, but I'm not sure I've taken any ISO 6400 images of a dark sky like this. You could certainly send the files to Nikon and ask...
     
  9. Banding is very obvious in the 1st picture. Less so in the other 2 pictures, but if you do a slight shadow/highlight adjustment in PS it becomes visible.
    You're not the first person to report this on the net, and there have been lengthy discussions about this. High ISO, high contrast - your shots are just about the ideal condition to produce this effect.
     
  10. I'm also using a calibrated everything, and can clearly see them in all three photos.
    Maybe you should send Nikon these photos and see what they say.
    I've done it before and they respond very quickly.
     
  11. The "banding" is light going into the night from floodlights. It is not the camera.
    Photogaphy involves the capture of light, and that is what your D700 (at ISO 6400) has done.
    A pure black sky would require the local electric company to cut power to the floodlights around the monument.
     
  12. Mike,
    There is some banding in your photos, this is due to high gain at the row sense amplifiers plus bleeding effect from the lights, I have noticed this in a few of my D700 shots, lifting exposures will cause the banding to be more noticeable. This is normal for D700 due to multiple channel readout configuration of the sensor that enables high speed read. some noise reduction software can correct this. Your camera is OK, remember this is ISO 6400.
     
  13. Yeh, it's normal considering the conditons: high ISO, low light with bright light sources in the frame. As Jerry noted, the banding is aligned with bright light sources in the frame. My D2H shows worse banding at 3200 and 6400, especially with any underexposure. Your photos are about a full stop or more underexposed. A lower ISO and/or less underexposure would minimize the problem. I never see any banding with my D2H between ISO 200-800, even with bright light sources in night photos.
    Check this site for descriptions and illustrations of different types of digital sensor noise.
     
  14. Thank you alll for your comments. From what I understand from all responses, this phenomenon is likely normal but, nevertheless, still disappointing. These examples are among the worst that I got but I would still like to send them to Nikon Canada (I am in Canada) to hear their opinion. Does any of you have an e-mail address where I should send them?
    By the way, the banding in this pictures is more evident on the back panel display of the camera than on my computer screen.
    Also, would I get the same banding problem if I were to use a lower ISO setting and longer exposure time, on a tripod?
     
  15. mjt

    mjt

    This isn't a side affect of the lights. When the D700 first came out (and to this day), the banding issue was discussed in the DPReview forums (same with the D3). Go to the DPReview D3-D1/D700 forum and type, "d700 banding" (no quotes). Here's a post going back to Sep 2008: http://forums.dpreview.com/forums/read.asp?forum=1021&message=29409559. The banding mostly occurs in very specific "photographic environments".
     
  16. mj t, the issue you're referring to, including the dpreview thread, actually confirms the observations made by Jerry, Arash and the link I provided which provides an overview of digital sensor noise. It is related to the presence of lights in the frame, which is clearly demonstrated by following the path of each segment of banding directly to a bright light source in the photos Michael attached here.
    As you said yourself, the "banding mostly occurs in very specific 'photographic environments'," which describes the circumstances illustrated in this thread: a very specific type of environment or condition, which includes operating at the outer margins of performance - high ISOs.
     
  17. Canon 5D MK11 manual expressly says that banding in the shadows may occur in very high ISOs with high contrast subjects - it'll be the same for the D700 (even though its the miracle camera!)
     
  18. ShunCheung

    ShunCheung Administrator

    The D700 may be a $3000 camera (at least the "list" price is $3000), but we still cannot expect miracles from it. Until a few years ago, ISO 6400 was mostly unhead of or of extremely poor quality. The D3/D700 lets you get decent/acceptable results at ISO 6400. However, when you have so many light sources inside the frame, there are some compromises. The $8000 D3X will give you noticably worse high ISO results; it is not intended to be a high-ISO camera.
     
  19. "Also, would I get the same banding problem if I were to use a lower ISO setting and longer exposure time, on a tripod?"
    no.
     
  20. The banding problem is (most likely) caused by the scarcity of bits in the shadow area of your image. There are many more bits available at the highlight end of the histogram; thus the axiom 'expose to the right'. If only a few bits are available to describe what is, in reality a gradient, it will stair-step, producing the result you see. This correlates with the presence of light sources in the image--it has 'fooled' your meter into underexposing.
     
  21. Mike,
    I have seen this issue at ISO 6400 only, not at ISO 3200 and lower, long or fast exposure. If you have Nikon Capture it is quite easy to correct, just drop a color control point over the banding area and reduce brightness, you will need one control point per band. You can also select-paint the whole sky and apply some Gaussian blur/darkening. Some noise reduction sw can also eliminate banding in a more sophisticated manner.
     
  22. I have never seen this with my D700 or D3 unless I accidentally underexpose the shot by a stop or more at ISO 6400 or higher, which is what you have done here. Price has nothing to do with this occurrence, it is user input that caused it.
     
  23. The "banding" is light going into the night from floodlights. It is not the camera.​
    No, it's definitely there in the dark areas of the first picture. I didn't see it at first, but now it is very obvious as vertical bands in the dark sky--when I am looking for it.
    Having seen it, however, I am not distracted by it. As others have said, it is understandable under these conditions.
    --Lannie
     
  24. i'm gonna disagree about the underexposure claim made here. i'd say that's pretty good exposure. if you'll look at the histogram, there is a peak very low, as to be expected from a shot of the night sky. however, if you look at the values of the monument, you'll see that the red channel is in areas very close, but not actually, to be clipped. this seems to have gained the most amount of light in its subject without clipping any of the channels, so i'm gonna argue that this image is basically perfectly exposed. the channels R and G are clipped in the point source lights but everywhere else they are not. so i argue that no more exposure could be given to this image without flat spotting one of the channels, usually that pesky red.
    now, as to the banding, were you shooting 12-bit raw?
     
  25. and was d-lighting on?
     
  26. ISO 6400 looked pretty good to me on the D700 (Active D-lighting ON set to AUTO):
    [​IMG]
     
  27. the reason that i ask is that on nx101.com, the author compares the 12bit and 14bit raw modes by underexposing by 8 stops a still life, and then seeing what he can pull back. in the 12bit, the picture comes back but with banding, the 14bit comes back without it and at higher quality. thus he finds that 14bit does hold more information and that there is some strange problem with banding in 12bit when pulled up by captureNX. this made me think that perhaps the camera tried to 'pull up' the exposure some on the night sky, thinking that it was underexposed, and thus encountered the banding problem. that's why i wondered if you were shooting 12 bit raw and had d-lighting on... if you'd like to read the page, it is
    http://nx101.com/12vs14.html
    and obviously
    nx101.com is the root and is quite an informative site. also, he takes a dig at ol kenny rock calling him an dolt. so maybe photo.net in general would like the site as well, n'est-ce pas?
     
  28. Thanks Dan for posting the link, but this does not apply to D700. Unlike D300, D700 uses native 14Bit ADCs while D300 has 12Bit ADCs. 14Bit mode in D300 is achieved by oversampling of the 12Bit ADC, data is read 4 times and therefore temporal difference between readout channels due to noise and cross talk is equalized by long-time averaging. However with D700 data is always 14Bit, when you select 12Bit NEF the last two bits are truncated when camera saves the NEF file. Bit depth has no effect on banding or noise in case of D700 and D3, however 14Bit can provide better protection against posterization in case of extreme tonal expansion in post processing of the NEF files and also provide less compression in subtle tones that result from minute differences in light intensity.
     
  29. Arash, do you know offhand whether there are more distinctly descriptive terms for the different types of "banding" that have been observed over the years with dSLRs? If so, it might help in clarifying these discussions. For example, the banding I see at high ISOs with my D2H, especially when underexposed, is different from the phenomenon that affected some D200's; and this "banding" demonstrated by the D700 is appears to be yet another type of artifact.
    Frankly, I don't really keep up with most of the technical aspects of Nikon dSLRs so I'm hoping someone with greater technical expertise might have some ideas about how to differentiate between these phenomena that are all lumped together as "banding."
    A further bit of confusion occurs when JPEG compression artifacts and posterization are also called "banding," but I doubt we'll be able to completely avoid that particular mistake.
     
  30. oh. ha. i took my stab at it. didnt think that dlighting would try to raise the exposure on the sky that much to do what he has found. but thats a good site in general if anyone wants to take a few minutes. it's nx101.com but the majority of the site seems to be well discussed and explained general information where captureNX information is just appended to a great compendium of other stuff. ha.
    does anyone else agree, though, that the picture is NOT underexposed?
     
  31. Dan, it's not an issue of whether the photos are underexposed in terms of aesthetic appeal. They look fine to me as-is. But just as with film, digital sensors have their limits. With film it's reciprocity failure. With digital it's noise at the outer limits of performance. With either medium we have to make adjustments to cope with those limitations.
    The type of banding I've described with my D2H (very different from the D700 banding illustrated here) can only be avoided by giving up to a full stop more exposure than accurate metering would indicate. The tradeoff is a loss of highlight detail. I just have to accept the limitations of a loss of dynamic range when pushing the D2H beyond ISO 800. But it's easier to tweak photos at ISO 1600-6400 to restore some semblance of normal tonal range than it is to try to edit out the banding artifacts.
    Another comparable situation is halation in film. I have some perfectly exposed photos with the camera aimed directly into the low angle sun. Even with anti-halation coating, the negatives show the effects of halation not only into the exposed frame but the outer margins and film rebate, where no light actually struck the film directly. I pushed the limits of performance for the film and that's what occurred. But the exposures are fine.
     
  32. Hi Lex,
    In a typical image sensors noise from photo sites is random and has a uniform distribution in the final image, any pattern noise including horizontal or vertical stripes which is referred to as "banding" is therefore associated with the readout circuit. High speed cameras like D700 have multiple readout channels (electrical pathways) to allow for rapid read and high speed continuous shooting operation. These channels conduct the analog signal from the sensor to the gain amplifiers where voltage is amplified and then passed to the ADCs (Analog to Digital Converter) for quantization. These channels and their corresponding amplifier stages are created equally, but there are minute differences due to the nature of the fabrication process. At high ISO and dark scenes like the example above, signal is weak and gain is very high, so the signal amplitude from each channel is slightly shifted due to these tiny mismatches. This will result in slightly darker and lighter stripes corresponding to each channel. This is of course a very simplified explanation, the onset of banding and in some cases the frequency of the stripes depends on the details of the sensor and peripheral circuit. Pattern noise is of course worse than random noise because it is visually more distracting.
    Regarding banding in D2H being different from D300, again it depends on the particular read out circuit that was used for D2H's sensor which I have little information about. One strange fact about D2H was that it was Nikon's trial to use JFETs for access transistor as opposed to regular MOSFETs, In practice this provided no advantage over Canon's superior CMOS chips and Nikon quietly pulled the plug on further development. Some of the artifacts unique to D2H are most likely results of the specific LBCAST sensor.
     
  33. These effects are not noise (random uncertainty in the data), rather they are artifacts (non-random distortions).
     
  34. Dan, I took the pictures in JPEG format, large, fine, optimal quality, with active D-lighting turned off. To be completely fair, the three photos that I posted are amongst the worst ones but, typically, all photos taken at ISO 6400 which have strong lights within the frame show banding to a certain extent. Sometimes this is barely noticeable, sometimes it is like in the photos above. However, I like taking night photos and I was hoping to be able to eliminate the need of a tripod with the D700, but it seems this is not the case. My biggest worry now is whether the camera is defective or this phenomenon is normal. From what I read between the lines, the latter may be the case.
     
  35. Thanks, Arash, very helpful info.
     
  36. nikon canada were extremely helpful when i had a problem with my d300 try rmak(a)nikon.ca
    good luck
     
  37. I have a more severe image banding problem with my D700. The cam was 3 weeks old when it 1st started to act up. Nikon Malaysia has agreed to replace my cam with a new one...even that, I had to fight for it. They said that my image picture circuit board (PCB) was faulty, at first they wanted to fix it since it was still under warranty. However, I demanded for a new one, after a few outbursts, they relented. As I was writing this, Nikon SC called to say that my new D700 is ready for collection :)
    00V2Mr-191819584.jpg
     
  38. Just an FYI, I just took a night photo and I definitely had banding.
    I noticed yellow banding when taking pictures of a white ceiling when I first got the camera, but I decided it was something that normally didn't come up. That was with my first camera. My second camera...well, that's the one I just took the night photo with and it has banding. Was this issue ever resolved for anyone else?
     
  39. I have been using my D300 for more than a year now. I have noticed something weird, lines (Banding) & green dots on black color started to come up. Reading thru this thread I've learn that it is beacuse of the faulty picture circuit board. Do you have some idea on how long will it take to repair and how much? thanks.
    00VsHP-224285684.jpg
     
  40. Henry, that definitely appears to be equipment failure. But none of us can tell you how much it will cost to repair or how long it will take. Only the Nikon distribution and repair center in your area can tell you that. I can only say that my one experience with Nikon service in the USA was positive. Warranty repairs to my D2H took only 2 or 3 weeks, from door to door, which I considered reasonable.
     
  41. Thank you Lex, I'm a bit disappointed knowing that the highly acclaimed NIKON D300 failed me. I honestly thought that this was “BUILT LIKE A TANK” although everything fail eventually but not just over a year. Adding to my frustrations is the fact that my unit cannot be repaired by our local distributor here in Qatar. It will be sent from Qatar to U.A.E. Dubai. I should have followed my friends advice CANON....
     
  42. That banding effect on my D700 is ruining most of my night shots - because light sources in the dark is the kind of night shots i like taking. After reading this thread, i will try higher exposures...
    However, I've had a Canon 5D II lent to me for the weekd-end, and it consistently showed less to none highlight banding. Here's an example. Was I very lucky with this Canon sample, or is this a general difference between models (or even brands) ?
    (note: the Canon 5D II's image was of course noisier when looked at at 100% magnification. But at small sizes, it looks much better. Strange... Btw, both pictures are RAWs developed with neutral settings)
     
  43. sorry. here's the illustration:
    00Xezv-300739584.jpg
     

Share This Page