2
\$\begingroup\$

Usually when I take photos I try to minimize the noise levels in the image.

However, what if I want to maximize the noise or generate an image of pure static, like the snow on a TV screen? Is there a way to do that?

Improve this question
\$\endgroup\$
6
  • 2
    \$\begingroup\$ Many image processing programs have a filter that adds noise, so there's no need to force your camera to do it. \$\endgroup\$
    – Blrfl
    Commented Dec 6, 2017 at 13:01
  • 1
    \$\begingroup\$ @ClicketyRicket is that important? What are you trying to achieve? \$\endgroup\$
    – osullic
    Commented Dec 6, 2017 at 13:46
  • 7
    \$\begingroup\$ @ClicketyRicket on the contrary, computer generated noise usually has better random properties than sensor noise because sensor is an imperfect physical device processing analog signals and tends to produce more or less visible patterns in the image \$\endgroup\$
    – szulat
    Commented Dec 6, 2017 at 13:50
  • 3
    \$\begingroup\$ @szulat nails it; see this question. Pseudorandom isn't a detriment unless there's truly-random noise is needed for cryptographic or statistical reasons, either of which would make this question off-topic. \$\endgroup\$
    – Blrfl
    Commented Dec 6, 2017 at 13:56
  • 2
    \$\begingroup\$ @ClicketyRicket If sensor noise were random from one exposure to the next, dark frame subtraction would be useless for noise reduction. \$\endgroup\$
    – Caleb
    Commented Dec 6, 2017 at 14:21

2 Answers 2

Reset to default
3
\$\begingroup\$

If you want the look of snow on a TV picture, then you don't want to do this in the camera. The snow is essentially random noise. Sensor noise is far from random. You can make the native camera noise a large fraction of the image signal easily enough, but you'll probably be disappointed by the nature of the noise.

The easiest way to make a snowy-noise picture is to add it in post-processing. Basically just add a random value to each pixel. For a little more sophistication, you can even adjust the frequency content of the noise.

By doing this in post-processing, you can chose the parameters of the noise arbitrarily, and try various configurations to see how they look.

As always, the best strategy is to use the camera to capture as much information about the scene as possible. That leaves maximum flexibility for post-processing, even if that processing includes adding noise or otherwise "degrading" image signal content.

Improve this answer
\$\endgroup\$
2
\$\begingroup\$

Is there a way to do that?

To maximize noise? Sure: set the camera to its highest ISO setting, and then take a photo with a very long exposure time. If you only want to see the noise, take that photo with the lens covered.

If you're looking to create a completely noisy image similar to a television tuned to a channel where there's no signal, you'll probably be disappointed. The static on a TV screen comes from the TV trying to interpret random bits of ambient radio energy as a television signal. The noise in your camera comes from the camera amplifying the signal it gets from the sensor, so that minute differences in value from one pixel to the next become noticeable. Those differences do increase with time as the sensor warms up, but at some point it probably levels off. So even though TV snow and sensor noise are both random, it's not surprising that they look different: they're different sources filtered through devices that process them differently.

Improve this answer
\$\endgroup\$
3
  • 2
    \$\begingroup\$ When I do this, the image tends to be pretty dark, not the gray I would expect. Do I just need to let it bake longer? \$\endgroup\$
    – Clickety Ricket
    Commented Dec 6, 2017 at 12:54
  • \$\begingroup\$ @ClicketyRicket Updated my answer ^^. I meant to address maximizing noise, not so much reproducing TV snow. \$\endgroup\$
    – Caleb
    Commented Dec 6, 2017 at 14:18
  • 2
    \$\begingroup\$ I think it is worth mentioning that you can get a lot of noise by a) long exposure or b) very short exposure and adjusting levels. In (a) case you add noise by increasing temperature of sensor and hence random thermal fluctuations. In case (b) you use fact that at low light level photon shot noise (en.wikipedia.org/wiki/Shot_noise) is higher (it scales as 1/√N, N=number of photons) \$\endgroup\$
    – aaaaa says reinstate Monica
    Commented Dec 6, 2017 at 14:34

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.