I created another test. This time I exposed for 36 minutes, because I wanted to test star photography. This is not meant to be a nice photo; I really just tested.

Camera was inside the house behind the window (it is too cold to go out) in dark. Minimal light from other room came from behind the camera. It is a Canon EOS D1000 set on BULB, F29, ISO800

The question is: Where do the blue and red dots (noise) come from?

sample image with noise

  • 1
    \$\begingroup\$ Did you have Custom Function CFn-3 (Long Exposure Noise Reduction) set to on or auto when you took this shot? This looks like long-exposure noise to me, which is caused by the sensor heating up due to the amount of time the shutter is open. \$\endgroup\$
    – Edd
    Jan 15, 2012 at 21:26

2 Answers 2


Well, perhaps you should have gone out after all :)

The noise is thermal noise, which will become noticeable as your sensor heats up during a long exposure. In astrophotography, it's quite a common problem.

Some ways to reduce such noise:

  • cool sensor down, e.g. by shooting in the cold weather. Note that cold also negatively affects battery life.
  • set the camera to take a dark frame with same settings (to get just the noise) and subtract it from the image; this can be either done manually (taking a second frame with lens cap on) or automatically by camera (the feature is called Long Exposure Noise Reduction or something similar).
  • using a stronger noise reduction setting in camera; this may also make the photo softer overall.
  • shoot several shorter exposures in a sequence and add them together in post processing (this is called "stacking"); you could use DeepSky Stacker for that. Make sure you do not combine this with dark frame technique to avoid dotted trails.
  • \$\begingroup\$ I would never thought about that. Really good thing to know. Thank you :) So I should go out with a AC adapter, so I have all the time in the world to shoot stars :D \$\endgroup\$ Jan 15, 2012 at 21:28
  • 2
    \$\begingroup\$ The Canon 1000D does have a Long Exposure Noise Reduction facility - it's under the menus as C.Fn-3 \$\endgroup\$
    – Edd
    Jan 15, 2012 at 22:27
  • 1
    \$\begingroup\$ Another method is to use image stacking technique, which is very popular when photograph star trails. You take multiple pictures of same scene and stack them up in post processing. \$\endgroup\$
    – Trav L
    Jan 16, 2012 at 5:36
  • \$\begingroup\$ @rockacola thanks for reminding, added stacking to list. \$\endgroup\$
    – Imre
    Jan 16, 2012 at 6:20
  • \$\begingroup\$ I've had a lot of success with the second option of taking a photo of just the noise. \$\endgroup\$ Jan 16, 2012 at 9:31

The Stanford CS178 lectures have a slide deck on Noise and ISO, which I shall attempt to paraphrase. There are several sources of noise, but the 3 biggest causes of noise in this photo are likely:

Photon shot noise Photons arrive probabilistically following a Poisson distribution. This means is there is a sqrt(intensity) variance in the recorded intensity of a light source. For darker images, sqrt(I) is very close to I, which causes significant noise. You can't really avoid this problem except by taking pictures of brighter things :)

Dark current Electrons in your camera's CMOS sensor are dislodged by thermal activity. Unless you live in a bucket of liquid nitrogen, it is too hot for you to avoid this type of noise. Going into the cold outdoors would reduce it. User Edd also points out that your camera has a Long Exposure Noise Reduction function, under C.Fn-3, which uses dark field subtraction in the camera to average out a lot of the thermal noise.

Read noise Heat causes the circuits that read out the data from your image sensor to record more noise. (This effect is reduced in CCD cameras.) Since you have a Canon, which uses CMOS, go outside into the cold.


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.