5
\$\begingroup\$

bad

Please don't mind the edited faces. All over the image are visible white stripes horizontally. What could cause them?

The image was taken without a flash on a Canon 5D Mark II. Could this be a sensor issue?

Shutter speed: 1/100
Aperture: F/2.8
ISO: 3600
Focal length: 24mm

The original image was in color and the stripes are colored aswell. Here is a part of the same image before any editing or exposure changes:

ex0

And here the same but the exposure increased by 1 EV:

ex1

Is this simply due to the rather high ISO or is there more? It looks a bit aggressive to me for 3200.

If it is indeed just caused by the high ISO (guessing the high voltage of the ADC is introducing the noise on the sensor?), I'd love an explanation on why the pattern is striped horizontally like that.

This is an image of the same camera with ISO 50 (Shutter 1/4, F/5.6) of the night sky:

ns

Looks pretty okay to me. Perhaps a bit too noisy for ISO 50 though?

Here is the same image but exposure raised by 1.22 EV:

ns2

No stripe pattern! And that is what's confusing me the most. How is the stripe pattern introduced?

Looking at this question and it's answer I start to doubt that this is only high-ISO introduced noise. The accepted answer there mentions how on a Canon 7D the pattern can appear vertical aswell. In the sample image in the question it is fairly equally horizontal and vertical I'd say, which both isn't the case for me. The stripes are rather clear (especially in the B/W image) horizontal lines, reminding me of the lines you'd see on really old TVs.

Improve this question
\$\endgroup\$
9
  • \$\begingroup\$ Guess: Sensor readout / gain control issues? Is the problem present with both camera-produced JPG and RAW? \$\endgroup\$
    – xiota
    May 13, 2019 at 13:53
  • \$\begingroup\$ @xiota I can't see a significant different between RAW and JPG, the images in my Q are from the RAW files. In the JPG there is a liiiiitle bit less noise visible, but I think that might just come from the JPG compression lowering overall image quality by a good bit, no big difference at all. At lower ISOs it does seem fine, if it were an actual issue with the sensor, would it just be present at higher ISOs? Maybe if someone with a 5D II could take a picture at my settings in a darker room to compare the noise performance. \$\endgroup\$
    – confetti
    May 13, 2019 at 14:42
  • \$\begingroup\$ photo.stackexchange.com/questions/41874/… \$\endgroup\$
    – Steven Kersting
    May 13, 2019 at 15:16
  • 3
    \$\begingroup\$ Possible duplicate of What causes banding noise? \$\endgroup\$
    – xiota
    May 13, 2019 at 19:16
  • 1
    \$\begingroup\$ @StevenKersting I'm not so sure... The example image there looks more like my ISO 50 shot, but not like my ISO 3200 shot. The answer mentions a 7D as example, talking about how the patterns might even appear more vertical than horizontal. In my ISO 3200 shot, I don't have any vertical bands. They look more like TV refresh lines you'd see on a very old TV, so I'm not sure if that really explains it and/or if that is all. \$\endgroup\$
    – confetti
    May 14, 2019 at 3:34

2 Answers 2

Reset to default
4
\$\begingroup\$

The most likely reason is LED event lighting (DMX cans, video projectors etc.) using fast and complicated modulation patterns for dimming or color choice. These units will not actually dim the light, instead switching it on an off at an extreme speed. Classic light dimmers work in a similar way, but incandescent lighting is too slow to follow the switching patterns. Inexpensive lighting kit seems to be worst, probably because you could improve the circuitry with an inductor - which however adds cost and weight.

This kind of kit will cause issues with any sensor that does not have all its pixels sampled at the exact same millisecond.

The mirrorball and PA equipment furthermore suggest a location where that kind of lighting gear is routinely encountered.

The exact effects are extremely dependent on camera model, shutter speed and shutter mode (eg Sony A7s in silent shutter mode is brutally susceptible to such!).

Improve this answer
\$\endgroup\$
3
  • \$\begingroup\$ While I'm not sure if that was the case (I was not the shooter, so I don't know all the details) it would certainly explain why the noise pattern looks so different in that shot than in any other shot I looked at and raised the exposure. What throws me off though is the perfect pattern. If these stripes were artifacts from LED lights, wouldn't it look more like random noise instead of a rather well aligned stripe pattern that repeats itself all over the frame evenly? \$\endgroup\$
    – confetti
    Aug 13, 2019 at 5:14
  • 1
    \$\begingroup\$ The dimming patterns LED equipment uses are in themselves not random. They would look very regular on an oscilloscope. A camera with a rolling shutter acts as an incidental oscilloscope more or less when forced to deal with it. Ever noticed the staccato pattern you see when quickly moving your head looking at some modern car's rear lights, or a typical LED clock? That is what your camera sees here. The technical term for these styles of modulation are PWM (pulse width modulation) when used for dimming, and multiplexing when used to save on wires. Or think morse code, rolling shutter as ticker. \$\endgroup\$
    – rackandboneman
    Aug 13, 2019 at 11:17
  • \$\begingroup\$ Thanks for the follow up, since it's definitely possible that a lot of LED strobes have been used there I will accept this answer as it seems to be the most plausible as in regular lighting even on high ISOs im not able to reproduce this pattern. Maybe I'll try with LED strobes in the future to see what will happen. \$\endgroup\$
    – confetti
    Aug 14, 2019 at 11:08
-4
\$\begingroup\$

I'm no expert at all, and as another said, I wouldn't know what is typical for a 5D2. But I see this is my Canon Ixus instruction manual:

"When you use flash photography at higher ISO speeds, the chances of white streaks appearing in the image increase the closer you approach to the subject". Not sure if relevant,

Improve this answer
\$\endgroup\$
1
  • 2
    \$\begingroup\$ As stated in my question, a flash has not been used. \$\endgroup\$
    – confetti
    Jul 22, 2019 at 19:31

Your Answer

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

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