3

I took two pictures with a Nikon D300 and a 17-55mm f/2.8G lens. Both pictures have been taken at ISO 200, aperture f/22 and shutter speed 1/8s. I used a tripod. The only difference between the two pictures is the framing (slightly shifted to the right).

Picture 1: enter image description here Picture 2: enter image description here

As you can see, the center of both images is smudged, whatever the framing. It is not a sharpness problem, the problem is about the center turning magenta/white.

Here are two zoom of the images for better viewing.

Picture 1 (zoom): enter image description here Picture 2 (zoom): enter image description here

Does anyone know the reason of the problem?

  • Very narrow apertures produce an effect called 'diffraction'. An excellent answer on the subject is here: photo.stackexchange.com/questions/8304/… – mpr Jan 12 '14 at 17:44
  • I'm pretty sure the problem is that you're expecting detail at a distance through air that's as wet as a very wet thing that has been left to soak for a while. You're getting very visible "atmospheric perspective" (reduced saturation and contrast, distinct blue tint) on the tree tops across the road; why would you expect small details to be sharp? – user2719 Jan 12 '14 at 18:40
  • I though the diffraction "only" resulted in less sharpness all over the image. Could the center turning magenta/white also be the result of diffraction? – Mia Xiu Jan 12 '14 at 18:41
  • 2
    Look more carefully. It's not the centre of the image, but the same patch of rock and vegetation that's showing the greying in both images (it's not in exactly the same location, it just happens to be near the centre in both), and whatever is in front of it is still green in both. – user2719 Jan 12 '14 at 18:52
  • 1
    It looks more like an optical issue to me as well. The spot is in exactly the same location (in terms of the sensor) in both photos. If you compare the color of the affected area in one photo to that same area (recorded by a different part of the sensor) in the scene in the other photo the color of the nearer objects is shifting as well. They're just darker and so the shift is not as noticeable. – Michael C Jan 12 '14 at 19:08
6

You can find a lot of others with this problem, especially some Nikon primes. Apparently, also zooms like yours. It is reflection from the glass in front of the sensor that reflects back to the lens rear element.

The colour and power of the spot depends on camera/lens combination, but often it is reddish (pink, magenta, whatever you want to call it) because some NIR-cut filters reflects a lot of it back.

(Filters are specified with transmission , absorption and reflection wavelengths. They do not cut 100%, and to cut a wavelength you can absorb it or reflect it. To absorb it absorbs some heat, so sometimes they choose to reflect instead. )

These reflections (both the NIR and visible reflections) can be reflected back into the sensor, When the surface normal points towards the sensor (which it does in the center!) thus increasing the power relative to the true signal. The true signal is pretty dim at F22, which is why you can see it at F22. Some users have reported red spots from F16, thus recommending to use minimum F11.

The reason is it not perfectly centered is because the true lens center and sensor center are not aligned perfectly, which is why camera calibration seeks to estimate the true center, and such a calibration can only be trusted for one focus distance and in case of zoom lens, one f-setting.

Hope this explanation wasnt too convoluted.

  • Great explanation, is there a word to name what you explain? Do you have technical documentation where I can have more information about what you describe? – Mia Xiu Jan 13 '14 at 10:41
  • havent found anything that focusses on this problem, but I see tidbits and casual observiations relating to it. See this one, where they confirm that e.g. Nikon D90 has a reflective NIR filter that causes trouble in astro photography, and you see the smudges in some of his photos. There are different ways to change the balance between the reflection and the true signal to make it visible. randombio.com/d90infrared.html – Michael Nielsen Jan 13 '14 at 13:45
  • note I say NIR and not IR because CCD/CMOS are not sensitive to IR, but to NIR they are. – Michael Nielsen Jan 13 '14 at 13:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.