I shot many long exposure photos (from 5 minutes to 30 minutes) back in December of 2010 with a two month-old Nikon D7000 and Tokina 11-16mm f/2.8 lens. The photos I took had very little noise. I used the same camera and lens last night (5 months later) and my photos have a ton of red, green and a few blue dots on them. My ISO was the same (200) and my aperture was almost the same.

So, what gives? Does noise increase as equipment ages? Or is there dust on my sensor? Or does temperature affect noise (ie, colder months have less noise and warmer months have more noise)?

Close up of the noise:

enter image description here


enter image description here


2 Answers 2


Inherently, no - time will not cause noise. You may have inadvertently damaged your sensor somehow if you tried to clean it manually or change a focusing screen or something internal with your camera.

I also doubt the ambient temp has risen enough to account for the difference, unless you were taking pictures in Alaska and you're now in a desert. But temperature will effect noise (you see this in astrophotography - but we're taking super long, repeated exposures, and even then its barely, if at all noticeable until we start messing with the picture in post).

Dust shouldn't show up as multi colored dots, so that's probably out.

My guess is that its you're photos last night were underexposed and they've been pushed too far in post and you're starting to see some noise.

Try posting examples and you'll get much more definitive answers.

Edit: OOOOooooOOOOO (after looking at your sample you've just posted) - thermal noise on the sensor from a long exposure. Is your camera's long exposure noise reduction feature turned on? That will help if its not on.

  • \$\begingroup\$ I did not use noise reduction as I was trying to take a series of photos, one after the other, so I could stack them in Photoshop to get a nice startrails photo. Not sure how to do in-camera noise reduction when trying to capture startrails. \$\endgroup\$
    – bperdue
    Commented May 21, 2011 at 16:07
  • 2
    \$\begingroup\$ Then you'll need to do dark frames and stack them in post. When you're done taking your pictures and before you pack up, put your lens cap on and take a series of pictures the same length and number of that are dark. These will JUST have the noise, and then stack them up and you can make a frame to subtract all the noise out. There's also specialized software for alot of this stuff like DeepStackStacker - but I'm not sure how it will play with star trails. Check takegreatpictures.com/photo-tips/digital-photography/… \$\endgroup\$
    – rfusca
    Commented May 21, 2011 at 16:13
  • \$\begingroup\$ Awesome, thanks so much for the help. Exactly what I needed. \$\endgroup\$
    – bperdue
    Commented May 21, 2011 at 16:24
  • \$\begingroup\$ The Startrails software (for Windows) is designed for this purpose, and supports dark frames. startrails.de/html/software.html \$\endgroup\$
    – coneslayer
    Commented May 21, 2011 at 19:58
  • \$\begingroup\$ An illustration of thermal noise appears at theory.uchicago.edu/~ejm/pix/20d/tests/noise/#thermalnoise . I don't think dark frame subtraction will help at all: these are random events, not consistent errors in specific pixels. But even extremely mild noise reduction algorithms ought to remove them. \$\endgroup\$
    – whuber
    Commented May 21, 2011 at 21:46

That looks like a few hot pixels my 20D has. Basically some of the photo sites in the sensor ramp up WAY faster than they should such that any exposure over a few 10ths of second and those subpixels go full on. The fact that those seem to be in 3 groups: 100% red, 100% blue and 50% green is a really good indication of that. I bet if you look at them in the RAW you'll see they're all exactly 1 px square. (They get distorted out by jpg compression.)


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.