1
\$\begingroup\$

I took a photo that has what seems like chromatic aberration. It's a purple color at the left and bottom edges, especially at the bottom-left corner:

enter image description here

This is a 30-second exposure at ISO 3200 with the lens wide open (F2.8) on the Sony NEX-5R. I then took another photo with the same parameters, but with long exposure noise reduction, and the fringe goes away:

enter image description here

To see the difference, please open both images in new tabs, full-screen your browser, and switch back and forth between the tabs repeatedly.

This seems like chromatic aberration [1], but then why does long exposure noise-reduction eliminate it? I thought long-exposure NR corrects for ISO noise, not CA, which is a property of the lens, not the sensor as with ISO noise.

Reducing the ISO to 400 gets rid of the purple color. There's still a white fringe, but it has lost its purple color. And it's less bright than at ISO 3200. Further reducing the ISO to 200 and then to 100 progressively reduces the brightness of the artifact.

So, am I correct that this is chromatic aberration? And how is it related to ISO and long-exposure NR?

Footnotes:

[1]: This purple fringe doesn't seem to be ISO noise, either, because ISO noise usually doesn't have a color, and doesn't occur along the edges of the photo.

\$\endgroup\$

1 Answer 1

2
\$\begingroup\$

There are two components of noise that most noise reduction algorithms reduce: luminance and chrominance. Luminance is the kind of noise that makes the shadows look grainy due to varying brightness from one pixel to the next. Chrominance is the kind of noise that causes spotty colors due to varying colors from one pixel to the next. Chrominance noise can be generated by long exposures as the sensor generates heat. Changing the ISO setting changes the amount of amplification applied to the analog information from the sensor as it is converted to digital data. The noise is also amplified along with the rest of the signal from the sensor.

The two examples in your question appear to exhibit a considerable amount of chrominance noise. The first seems to have it worse in the lower left corner, but the second has it worse in the middle of the light colored wall. This is what you seem to be mistaking for chromatic aberration.

\$\endgroup\$
4
  • \$\begingroup\$ Thanks for the link, great read for a photo-newbie like myself. I'm not convinced by your conclusion, though. The purple glow doesn't look like chroma noise to me. I wouldn't call it noise at all, as it doesn't seem to have the spatial nor the amplitude distribution of your typical thermal noise. \$\endgroup\$ Dec 19, 2015 at 12:07
  • \$\begingroup\$ Could it also be a colour cast? Light reflected from other coloured objects in the environment? The only difference I see in those 2 pictures are the pattern of the noise, but no chromatic aberration. \$\endgroup\$
    – Dragos
    Dec 19, 2015 at 12:38
  • 1
    \$\begingroup\$ Chrominance noise is not the same thing as chromatic aberration. Far from it. \$\endgroup\$
    – Michael C
    Dec 19, 2015 at 15:40
  • \$\begingroup\$ @Dragos Not being familiar with that particular camera It is hard to say for sure regarding the left edge of the first photo. But there are some cameras with sensors that demonstrate that kind of readout noise when used for long exposures. \$\endgroup\$
    – Michael C
    Dec 19, 2015 at 21:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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