For night photography, do long exposures work better or worse than taking a few photos and merging them in software (median blending / image stacking, like the excellent iOS app Cortex Camera)?

You can assume that the images are perfectly aligned -- assume the blending software corrects for both horizontal and vertical shifts and slight rotations. $2.99 iPhone apps do this, so it's not an unreasonable assumption. Alternatively, assume that a tripod + remote shutter release are being used, to eliminate movement between the photos. Long exposures use a tripod, of course.

So, does median blending result in better or worse image quality? Needless to say, assume other parameters (lens, aperture, ISO, camera...) are all the same.

(This is split from Can I merge multiple photos to reduce noise? but that question specifically assumes that I don't have a tripod, whereas this one allows for it. So they are different questions. I went through the links suggested there, but they didn't seem to clearly answer this question.)

3 Answers 3


Theoretically the mean of 10 one-second exposures should give the same amount of noise as one ten-second exposure.

The results in practice differ mainly on account of thermal noise. The longer the sensor is active during an exposure the warmer it gets which results in an increase in dark current noise. Multiple short exposures allow the sensor to cool in between. There is a threshold where thermal noise starts to become a real issue (I don't have the data to hand but it has been studied at length by astrophotographers).

Shorter exposures are also favored if one wants to avoid star trailing.

One advantage of multiple short exposures is that you have the choice of averaging methods, a long exposure can only combine photons in an additive way, whereas with multiple exposures you may capture the same amount of photons in total, you have extra information of when the photons arrived.

Certain types of noise will respond to certain methods, mean and median are usually pretty good and widely implemented. Better still would be a hybrid method like the alhpa-trimmed (truncated) mean. This discards outliers at either end that can skew the result and then computes the mean. You could do even better but that would involve profiling and calculating the parameters of the individual noise distributions.

However as the number of exposures increases the results will converge, i.e. if you shoot enough exposures it will cease to make a difference whether you use mean or median or some other method.

  • Matt, you are assuming all the exposures are added to each other. Median blending averages out the noise, which allows you to take a series of shorter exposures at higher ISO instead of a longer one at lower ISO.
    – Michael C
    Nov 26, 2013 at 22:32
  • The bottom image is a Median Blended version of the other five. Notice there are no cumulative artifacts caused by the object moved across the FoV in each of the first five photos. The software compares the five photos and throws out the anomalus values from one frame rather than simply averaging the results of all five frames. petapixel.com/assets/uploads/2013/05/ultramarine-combined1.jpg
    – Michael C
    Nov 26, 2013 at 22:48
  • 1
    Matt, you said that theoretically the MEAN of 10 one-second exposures should give the same amount of noise as one ten-second exposure. Agreed. But shouldn't the MEDIAN of 10 one-second exposures give less noise? Median(1, 1, 1.1) = 1, while Mean(1, 1, 1.1) = 1.033. Nov 27, 2013 at 1:46
  • @KartickVaddadi I'm aware of the difference between median and mean, I said mean for the sake of correctness, if the noise follows a normal distribution then the mean is the best estimate of the true value, however photon noise is not normally distributed (it follows the poission distribution), and there are other noise sources. Discussion of the difference in effectiveness between mean and median for the sum of a series of unknown distributions belongs on math.stackexchange.com, however in all likelihood the results would be similar (for a static scene).
    – Matt Grum
    Nov 27, 2013 at 9:19
  • 1
    @MattGrum At higher ISOs the benefit of median over mean is much more apparent.
    – Michael C
    Nov 27, 2013 at 10:47

From my early experiments with astro photography (actually using a special webcam) I can tell that it is the usual way to make images there to make a video at 25fps and then stacking the images with dedicates software. The reasons for that are exactly that the noise from the atmosphere can be factored out, and even a quite unstable tracking system can be accounted fore in software afterwards. Sensor cooling could not be a topic there because the sensor is active the whole time (several minutes usually).

This might not be a satisfactory answer for your exact question but in my eyes it points towards multiple images might yield better quality, provided the stacking software is good enough.

  • Interesting idea. Is there easy-to-use software available for the Mac: just one-click -- you drag a video in and say Save As PNG or something. Note that I use a Mac, and don't want command-line software, or expensive stuff (like $50). What do you think about shooting a video vs shooting a burst of images? The latter will have higher resolution. Cortex Camera, in fact, shoots a burst of photos on one iDevice I tried it on, and a short video on another. I don't know why it does that. Nov 27, 2013 at 1:29
  • I also use Macs. The software I used for stacking was Windows only, I am afraid. And I don't know of any Mac OS alternatives.
    – Frank
    Nov 27, 2013 at 14:49

The blending using several layers of little to no noise will be better than a single exposure with lots of noise.

A great write up of median stacking (which you specifically asked about) takes the median value of each pixel from each layer - is done here:


It also compares visually both methods. Definitely worth checking out.

Other methods: Software blending, for example, can automatically use the contents of that layer only if the level of brightness is at a certain level. This reduces dark noise that exists with a single exposure. This is often seen in the lighten layer blending mode in Photoshop. Just another way of going about it.

Best of luck, and have fun.

  • Thanks, this answers my question. A nit: when you say "takes the average value", don't you mean "takes the median value"? Taking the average value will still keep some of the noise from the original shots, whereas the median completely eliminates that noise. Median(1, 1, 1.1) = 1, while average(1, 1, 1.1) = 1.033. Nov 27, 2013 at 1:41
  • -1 The fact that blending several layers with little noise is better than a single exposure with lots of noise is blindingly obvious, the question asks whether blending several layers is better than a long exposure, the assumption being that the total exposure time is the same in both cases. Neither this answer or your link address this point at all.
    – Matt Grum
    Nov 27, 2013 at 9:29
  • @KartickVaddadi median doesn't completely eliminate noise at all, depending on the noise distribution it can be worse than the mean: mean(1.01 0.95 1.03 1.02) = 1.005, while median(1.01 0.95 1.03 1.02) = 1.015
    – Matt Grum
    Nov 27, 2013 at 9:37
  • Point taken. The petapixel link was still interesting because it says that image stacking helps even at base ISO. But you're right that the exposures are different in that case -- we're comparing a photo at a certain exposure with a bunch of photos at the same exposure. Whereas my question asks about long exposures. Let's discuss this below, in the comments on your answer. Nov 27, 2013 at 12:38
  • @matt Grum Why on earth would you use the same exposure time? That defeats the purpose. There is nothing in the question at this time by the OP that states that. The questions asks about usage, not theory.
    – Cymbals
    Nov 29, 2013 at 15:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.