Pages: [1]   Go Down

Author Topic: Long exposures & noise  (Read 2848 times)

Roeland

  • Newbie
  • *
  • Offline Offline
  • Posts: 9
    • http://
Long exposures & noise
« on: December 17, 2008, 02:28:49 pm »

Hi everyone,

During my last fotoshoot in Sweden, when photographing just after sunset, exposure times became as long as about 30 s. Being a 5D user I can't help being a purist for low noise so I kept it down at ISO 100, hence the long exposure time. Back at home I discovered horizontal and vertical noise patterns in the dark areas when zooming in on the image. This effect is  very minor and far from visible when viewing the image set at a normal size. However, would it be possible to have less noise by setting the ISO at let's say 400 and shooting at a shorter exposure time? When is the moment to decide to shorten the exposure and choosing for a higher ISO?

Roeland
Logged

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Long exposures & noise
« Reply #1 on: December 17, 2008, 03:14:22 pm »

Quote from: Roeland
I discovered horizontal and vertical noise patterns in the dark areas when zooming in on the image
That pattern in inherent to the sensor and the way Canon cameras record the raw data (the same if it gets converted in JPEG in-camera), but normally it "falls under" the black level, i.e. it vanishes in the darkness. When you have to increase the intensity (for example by the "exposure" slider in ACR/LR), the pattern appears.

The appearance of the pattern depends on the way raw the raw processor interprets the data: it may be dominantly horizontal or cross like.

See the attached images; the first one is from an ISO 3200 shot of Ray; it demonstrates, that the pattern depends on the intensity. The other is from an ISO 100 shot of a black sheet, the raw channels are shown separatedly, you can follow the pattern. This shows, that the pattern does not depend on the ISO but on the pixel intensity.

So, plainly, it is the too low exposure. If there is no danger of clipping of important details, use higher ISO, not instead of the long exposure but in addition.
Logged
Gabor

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Long exposures & noise
« Reply #2 on: December 17, 2008, 03:38:30 pm »

Here are four slightly different interpretations of an ISO 100 image of Ray (shot for DR measurement, it is extremely dark, requiring +6 EV). The point is, that the appearance depends on the raw processor.
« Last Edit: December 17, 2008, 03:40:35 pm by Panopeeper »
Logged
Gabor

Roeland

  • Newbie
  • *
  • Offline Offline
  • Posts: 9
    • http://
Long exposures & noise
« Reply #3 on: December 17, 2008, 04:22:25 pm »

Thanks for taking the time to reply,
you clearly have wandered beyond the scope of your camera manual!
So if I got this right, the noise patterns I have are a fixed quantity for my specific equipment configuration which I can have outnumbered by more light hitting the sensor (signal to noise ratio)?
I have heard of random 'photon shot noise' and also 'read noise', but until now I never thought about how my RAW data gets processed.

Time to improve my knowledge about this matter...



Logged

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Long exposures & noise
« Reply #4 on: December 17, 2008, 05:42:38 pm »

Quote from: Roeland
you clearly have wandered beyond the scope of your camera manual!
LOL, I don't have a 5D but a 20D and a 40D. However, I am the author of Rawnalyze, the program which helps analyzing the raw data; as such, I have collected any analyzed literally thousands of raw files from dozens of cameras.

Quote
if I got this right, the noise patterns I have are a fixed quantity for my specific equipment configuration which I can have outnumbered by more light hitting the sensor (signal to noise ratio)?
Almost right. The pattern is not SNR dependent; in that range, virtuallly there is no signal, but that does not appear so clear-cut. I would say the pattern is the manifestation of the lack of signal. "Zero signal" is not known in Canon raw files (well, except ...) and the processing program has to decide, what it regards as zero, i.e. total black.

Quote
I have heard of random 'photon shot noise'
That becomes a factor only with high exposure.

Quote
Time to improve my knowledge about this matter...
Upload one of your raw files as sample for an analysis to gain a deeper understanding. Photographers are often surprized to see the truth about their shots (some don't even want to see and accept the truth).
Logged
Gabor
Pages: [1]   Go Up