5

I'm taking photos at night with a GH1 at ISO 640 and getting a ton of noise in the dark areas.

I've seen a bunch of other peoples night time photos and they never have this much noise. I tried using a really long exposure of 25 seconds to compensate for the low light but there's still a ton of noise. Is this unavoidable or are there other settings I can use to get rid of the noise? Or am I supposed to de-noise all the images in post?

The first shot is right before the moon came out and the second shot is when the moon was fully out (its was around a 65% full moon that night). I also shot these in Aperture Priority mode if that makes any difference. Thanks for the help!!

EDIT: I was googling and I just came across Exposure to the right(ETTR). Could this be what I'm experiencing here? The images were underexposed and therefore lose a lot of values?

20mm, F11, ISO640, 25sec exposure

20mm, F6.3, ISO320, 15sec exposure enter image description here

4
  • If you are underexposing and pushing the exposure in post processing, this will bring out more noise then you would otherwise see.
    – dpollitt
    Feb 2, 2012 at 20:43
  • 1
    Hi thanks for the info as far as that goes, just to be clear about the above images, they are right out of the camera with no post processing done. Feb 2, 2012 at 20:55
  • You should definitely expose to the right in that situation, and open the aperture right up, at 20mm depth of field should not be a problem.
    – Matt Grum
    Feb 3, 2012 at 16:10
  • The most significant difference between these two pictures is the difference in color temperature/white balance between the two.
    – Michael C
    Oct 20, 2013 at 15:38

3 Answers 3

7

This question is probably best answered in two parts.

Part1:

You may need to increase the ISO to combat noise. This sounds counter intuitive due to common misconceptions regarding noise.

Noise is principally caused by lack of light. Lightsources emit photons randomly, the more photons you collect, the more the randomness averages out, leaving a smooth image. Fewer photons -> more random -> lots of grain.

All increasing the ISO does, is amplify the signal (and noise) you get from the sensor prior to digitization. It doesn't create noise, and as it amplifies the signal and noise by the same amount the signal to noise ratio remains the same. Needing to amplify the signal implies you have a weak (and hence noisy) signal to begin with, hence the association with high ISO and noise.

Noise is also introduced when trying to read the analogue signal. By amplifying before reading you add a small amount of read noise to a big signal, with little effect. By shooting at a lower ISO (keeping the other settings the same) you amplify less, to you add a small amount of read noise to a small signal, giving you a worse result by lowering ISO.

Part 2:

Reducing read noise by upping the ISO will help, but the primary problem you have is lack of light. You need to get more light down the lens, by either opening the aperture, or increasing the exposure time.

If increasing the exposure time/aperture causes other parts of the scene to be too bright (to the point where they show up pure white) then you have an additional problem that the dynamic range (the difference between the brightest and darkest parts) of the scene is too high, this can be fixed by using graduated filters, or by combining multiple exposures.

If you have a static(ish) scene I would recommend shooting several identical 25s exposures and then averaging them in software to reduce the noise. This is analogous to having one very long exposure, except without problems due to overexposure of certain areas. This is how most good DSLR astrophotography is done.

1
  • 2
    Need to back Matt up here. In your case @trying_hal9000, your very probably using a far lower ISO setting than you need to. The goal is to saturate your image during the exposure...usually you can do that with a longer shutter or wider aperture...but when you don't have the option of changing those, you really need to increase ISO, sometimes by a fair amount.
    – jrista
    Feb 3, 2012 at 16:19
4

It isn't unusual to get noise in shadow areas even at low ISO.

What is the native ISO of your camera? There are theories that you want multiples of that native ISO. So 100, 200, 400, 800 rather than intermediate values like 320, 640. I'm not sure if that's established fact or old wive's tale.

You camera may have a long exposure noise reduction feature. This does a "dark frame" to capture any noise caused by the camera internals, and will subtract that out.

Failing that, you can take multiple images and stack them and have software average out the dark areas. Oloneo does this for example, or if you use Photoshop you can load the images into a layer stack and apply a layer stack mode of median.

3
  • All long exposure noise reduction captures multiple frames to cancel out noise?! I had no idea!
    – dpollitt
    Feb 2, 2012 at 20:41
  • My camera is a Panasonic GH1 and from what I've read people think it's native ISO is either 320 or 640. I've read that Canons native ISO's are actually multiples of 160 too and not the normal 100 multiples. But I find that the info seems to always be conflicting as far as determining a camera's native ISO is concerned. I will try out the camera long exposure feature and see how that works. Feb 2, 2012 at 21:03
  • What I think you are referring to is the fact that Canon DSLRs implement third stop ISOs 320, 640, 1000 etc. in software rather than hardware, which has a small but measurable effect on noise. For that reason I stick to the powers of 2, 200, 400, 800 etc. I don't know if this effects other manufacturers, but it's worth repeating that the effect is small. It has nothing to do with native ISO.
    – Matt Grum
    Feb 3, 2012 at 16:08
2

According to this article, the Panasonic Lumix DMC-GH1 has a long shutter noise reduction on/off setting.

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.