Jump to content

Stripes - how to correct?


kari_oinonen

Recommended Posts

<p>I took some photos in dark. I can see a street light making a vertical stripe or a band. These are seeable also in the small picture I try to attach.<br>

Equipment: d700, Ais-G 28mm/f2.8 Nikkor - chipped manual focus lens. I admit that I was also testing the high ISO settings. This time I was running at 6400.<br>

Clearly this is not a lens feature as two different lights make perfectly vertical band. I understand that the situation was tough for the sensor.<br>

Do you have a good explanation on this? Is it a owerflow or what?<br>

Is this issue possible to correct afterwards easily?<br>

Cheers Kari</p><div>00ZQzO-404671584.JPG.8b3c9f64731f1a21c9fa6836aad9a9f9.JPG</div>

Link to comment
Share on other sites

<p>Hi Kari,<br>

I would say this scenery is simply a bit too hard for the D700's sensor. I shoot the D700 a lot at higher ISO-settings but not extreme-contrast-scenes like the one above. I'm not aware of an easy correction.<br>

My first DSLR (Nikon D1) was showing banding at much lower ISO-settings (it could be deliberately provoked at ISO 400). <br>

Cheers, Georg!</p>

Link to comment
Share on other sites

Link to comment
Share on other sites

<p>Thank you.<br /> Now I know it is called sensor blooming. The link and explanations helped. I was, correctly, thinking that it is mostly sensor related.<br>

In the attached picture the exposure was 1/50s and f5,6.<br>

<br /> So is it correct to suppose that only post processing can help in the case blooming exists?<br /> Are the corrections doable in CaptureNX2? I do not have that yet.<br /> Cheers Kari</p>

Link to comment
Share on other sites

<p>I've used my D700 quite a lot in very low light and never noticed anything like that shown above. Even maxed-out at 25,600 EI and with stage spotlights in the lens there's never been any sign of sensor streaking.</p>

<p>There is one optical circumstance that could cause the effect without it being the fault of the camera or sensor. That's if the lens has a horizontal smudge across it. Unlikely I know, but still a possibility. So check out the lens and make sure it's immaculately clean and totally free from grease and smudges. If the same thing is still happening then, I think you should return the camera to Nikon, because that really isn't typical of a D700, IME.</p>

Link to comment
Share on other sites

<p>Joe, jpegs are fine at least, there is a max size limit too.<br>

Started thinking again, based on that you did not have blooming in your d700. - Could it be that these street lights spectrum has anything to do with this? In the neighbourhood the lights have been exchanged to a low energy type recently.<br>

The lens used is some 20+ years old. It has got minor wounds, hardly detectable. Otherwise my lenses are very clean.<br>

Maybe I do a little it of more testing to see when and where this supposed blooming comes.</p>

Link to comment
Share on other sites

<p>I do not see a <em>vertical</em> band. I do see a star pattern emanating from the street light. Star patterns result from a narrow aperture, perhaps F8 or F11 or narrower. Some photographers use small apertures like that precisely to obtain the star pattern. EXIF data could not be extracted on my end from this image for aperture, etc.</p>

<p>I do see what could be <em>horizontal</em> bands. When I jack up the mid tone or the shadow contrast to the extreme, they are visible on my NEC in CNX2. When contrast is reduced, they are not visible. So, I vote for sensor blooming, not banding. This seems to occur only at very high ISO, when there is also very high contrast. Curious. I have never seen this in my own D700.</p>

<p> </p>

Link to comment
Share on other sites

<p>To make these effects more obvious, I've brightened up the image - see attached.</p>

<p>The spokes of light radiating from the street light are due to diffraction from the edges of the partially closed aperture in the lens. </p>

<p>WRT banding, I don't see much of this in the vertical direction, however there is appreciable banding in the horizontal direction. This effect used to be quite common in early CCD sensors, ie, http://www.andor.com/learning/digital_cameras/?docid=395, and http://en.wikipedia.org/wiki/Charge-coupled_device. It was sometimes called "sensor smear" or "bloom". It's less common in modern CMOS sensors, but can still occur. In the attached image, I've indicated the major horizontal bands with thin red rectangles with rounded corners.</p>

<p>Even in photoshop, it can be quite tedious to get rid of this sort of problem manually, but there are plugins which make short work of it. Attached to the next post is the result of running the brightened image through Topaz De-Noise, using it's de-banding option. If you don't have this particular plugin, the next best bet is to use one of the free Fourier Transform plug-ins and handle it in the Fourier domain.</p>

<p>HTH,</p>

<p>Tom M</p>

Link to comment
Share on other sites

<p>Strange ... I can't attache a simple jpg ... let me try again. Sorry, something must be up. I've uploaded hundreds of jpg images to this forum and it won't let me do it, saying that it doesn't support jpg file type? ! ? ! ! !</p>

<p>I'll try again in a little while. Sorry.</p>

<p>Tom M</p>

Link to comment
Share on other sites

<p>Thanks for the info., Tom. I've not encountered the problem before but good to know your suggestion on how to get rid of it. </p>

<p>I don't know the gain structure of the D700 but presumably ISO-6400 is still in the range of hardware gain, in which case the degree of overload of the bright light isn't surprising compared to properly exposing it by spot metering. </p>

<p>Reducing ISO will also reduce blooming, but I wonder if making up for the underexpose in post-processing will significantly reduce signal to noise ratio; I'm guessing probably not. </p>

 

Link to comment
Share on other sites

<p>BTW, Rodeo, you mentioned the possibility of the horizontal bands being due to scattering from a horizontal smudge. Actually, if it was due to scattering, it would have to come from a thin, vertical straight-line scratch. A line scratch scatters in the direction perpendicular to the scratch, and the thinner the scratch, the greater the angle over which it scatters.</p>

<p>Michael, unfortunately, I don't know have a definitive answer to your questions about gain structure, recovering brightness in PP, etc. Like you, I would presume that such a high ISO would be in the range of hardware gain. If so, from the way the horizontal band overlays on the darker parts of the image, I would guess the amp(s) are working linearly. If so, I would guess that it wouldn't make much difference in the strength of that streak if you set the camera for ISO 6400 (as done by the OP), or 12800 and didn't use as much PP brightening. </p>

<p>With respect to having observed this in my two d700s, I don't tend to take shots like this (ie, at night with a very bright light in the frame), so I could only find one in my archives that was vaguely similar, but it was at ISO 2500. However, when I brightened mine up by a similar amount, I can start to see a horizontal band centered on the bright light in my image, just like in the OP's image. This is hardly a good example/comparison because I was just playing around, not doing a serious test. For example, I was almost certainly shooting with a different lens (the relatively new 16-35 f4, in my case), had left the UV filter on, and was shooting through the motel room window.</p>

<p>Tom M</p><div>00ZRKm-404953584.jpg.b35b7c78ca12948ffdbd1ed7119cdfda.jpg</div>

Link to comment
Share on other sites

<p>I admit, my error. I was just 90 degrees off. The aim was for the term horizontal, as you have noticed.<br>

Tom's photo attachment shows the same horizontal band.<br>

I did some tests, indoors this time. At the back of the room there is a normal flashlight.<br>

In the first I was using ISO 6400.</p>

<div>00ZRN0-404995684.JPG.4d039cd8c76c74870ab77241d61e1ae8.JPG</div>

Link to comment
Share on other sites

<p>As you can see the blooming band is there in Indoors 1. For the next photo I tried to keep exposure the same but changed ISO to 1600. This time I can not see banding. So I had to reduce the sensitivity by two stops to avoid the problem in this indoors experiment.</p>

<p>So maybe you can not tell exactly the ISO value where the band does not emerge. Seems that the appearence of the band is dependent on several factors like ISO and the relative brightness of the light source and probably some additional factors.</p>

<p>Because of not using a support, the Indoors 2 photo shows some movement. I believe that it is not so important in this case.</p>

<p>It is good to know that a treatment like Topaz exists, in the case of need.<br>

Cheers Kari</p><div>00ZRNB-404997584.JPG.4f04f9d9acb614f1d483c22ffb3bdc85.JPG</div>

Link to comment
Share on other sites

<p>Yes, that's "Sensor Bloom" and it can happen when you've got the ISO cranked up in dark scenes. My advice for this type of photographic situation is to shoot some film. Preferably Kodak Ektar 100 or Portra 400. You certainly won't see any Sensor related issues, and 35mm Nikon film bodies are dirt cheap right now. You already have the manual focus lens for a Nikon FM or FM2.</p>

 

Link to comment
Share on other sites

<p>It only happens at high ISO *AND* you have a scene which is mostly dark but includes some extremely bright light sources, ie incredible contrast ratios. </p>

<p>Film is not the answer if you need really high ISOs (because of subject motion or other considerations). I don't know of any color film that could possibly have produced the first two example images in this thread, ie, the first shot at 6500 and mine at 2500 (but cranked up in PP to somewhere around the same ISO or higher). Obviously these two examples didn't require such high ISOs, but there is plenty of night street, night sports and other kinds of photography that need such values.</p>

<p>Tom M</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...