24

I am doing some research to get into astrophotography.

Most of tutorials recommend taking multiple exposures and then stacking them.

That makes me wonder: provided that the camera is on a tracking system, why not leave the shutter open for as long as possible?

  • 2
    Do these tutorials assume you have a tracking system in place? – unperson325680 Jan 29 '18 at 8:49
  • 1
    I recommend the book "Deep-sky Imaging Primer" by Charles Bracken. It will answer most beginers questions. For more details refer to a dedicated astro forum like cloudy nights. – user71927 Jan 29 '18 at 12:23
  • 2
    Related on Physics: physics.stackexchange.com/questions/306036/… – dmckee Jan 29 '18 at 23:15
43

@Michael Clark and @Itai have provided good answers. A few more thoughts from the perspective of the enthusiastic amateur:

  • Tracking technology isn't perfect and sometimes its better to work within the practical limitations of the tracking available rather than push it too far

  • Very long exposures may not play well with high levels of light pollution. There's a balance to strike between catching as many target photons as possible without overdoing the unwanted ones

  • Risk management: if you have just one long exposure then it can easily be ruined by a flash of external light going off. Multiple, shorter exposures mean that you can just throw away the bad frames

  • 9
    Re " There's a balance to strike between catching as many target photons as possible without overdoing the unwanted ones": That should be utterly independent of exposure time. – Peter - Reinstate Monica Jan 29 '18 at 9:48
  • 2
    +1 for the tracking technology comment, with a dumb driven polar aligned equitorial I've been able to get ~3 minutes at 300mm FL without star trails before now, but only with luck on my side. Depending on the quality and sofistication of your kit and technique, you might do better or worse, but aligning and stacking in post will always help if you're trying for deep sky objects with amateur kit. – Joseph Rogers Jan 29 '18 at 10:49
  • 2
    @PeterA.Schneider There's time-varying interferers like meteors, airplanes and clouds. Taking multiple images lets you easily discard these. Another factor is saturating the photon storage of a pixel, a limit called "full well capacity". – user71659 Jan 30 '18 at 4:55
  • 3
    @PeterA.Schneider That's not actually true unless the "unwanted light" is near constant AND ubiquitous. Or you have no algorithm in place for discarding the bad frames. If you have many short exposures you have many samplings from some distribution. The stars stay in (nearly) the same place (assuming a tracking system) the unwanted light is likely to be either on off (cars, houses, even street lights these days) or no stationary with regards to the tracking reference frame. Both of these let you filter it with smartish algorithms. – DRF Jan 30 '18 at 7:58
  • 1
    @DFR you say "discard bad frames" and I can't tell if you are agreeing with MartinV's point #3 or if you are implying a digital composition of multiple sequential images into a simulated long exposure (adding the word algorithm I think the latter). I don't know about the OP, but my long exposure [very amateur] astrophotography is done with heavy glass on honest-to-gosh film. You can't assume the availability of digital cleanup processes in long-exposure unless the OP explicitly says it is there. – Ruscal Jan 30 '18 at 23:42
18

It is firstly because we can now.

Bulb photography can indeed shoot exposures of minutes to several hours, depending on the camera. Using a film camera, astrophotography is done with very long exposures and those cameras have no time limit since they do not need power to operate.

A digital camera can be used in the same way but most mirrorless limit bulb exposures to 30 minutes or an hour, making it impossible to shoot a longer exposure, so there is no choice.

Most DSLRs though can take exposures of over an hour, so they can make the very long exposure to get the final image in one shot. However, using multiple shots can be advantageous. Most importantly, the maximum brightness of each pixel is virtallu expanded. In one shot, once a photosite is saturated, it will be over exposed. With multiple shots, it is possible the saturation will not occur, giving software more data precision to work with. Grosely over-simplifying, each doubling of shots give you an extra bit of precision and dynamic-range roughly. So take just 4 shots, gets you 2 more stops of dynamic-range compared to a single exposure.

Multiple exposures have the effect of averaging noise. This will give you cleaner images to work with but each image can have software noise-reduction applied which is more effective than processing a longer much noisier exposure.

It is rather hard to meter for a very long exposure but if you have many images, you have flexibility to not use the whole stack or control software blending to avoid over-exposure, particularly if you have included some foreground in your composition.

The two disadvantages of doing multi-image capture are minor. One is that it is more work later since the stack of image must be transferred and processed by computer, instead of a single image. The other is that there may be tiny gaps if you are doing star-trails while the camera takes time between shots (make sure to turn off Long Shutter Noise Reduction if you do nou want huge gaps) which can depend on a number of things and the particular camera.

18

The main advantage of stacking is to average out the randomized Poisson distribution "shot noise" that can be a problem in low light images such as astrophotography. Another advantage for stacking comes in using dedicated monochrome imaging sensors while alternating color (or specialized astronomy related) filters over the entire sensor for each exposure and then combining them in post.

  • 7
    for shot noise stacking multiple exposures is exactly the same as using a single long exposure of the same accumulated time. we can think of it as "analog stacking". so, no, this is definitely not the reason why we use multiple exposures. – szulat Jan 29 '18 at 11:16
  • 6
    @szulat Only if you stack the multiple exposures using plain average filtering, as opposed to trimmed average or median filter. – jpa Jan 29 '18 at 11:27
  • 1
    @jpa the answer states "average out". anyway, algorithms are secondary. the process itself gathers the same amount of noise. what can be different is technical, e.g. the ability to obtain much bigger dynamic range (than a single exposure. 14 or even 16 bit files are often not enough to cover enormous contrasts of astrophotography) – szulat Jan 29 '18 at 11:46
  • 4
    @szulat: Remember - the median is a type of average. – psmears Jan 29 '18 at 12:47
  • 2
    @MichaelClark in science, these techniques are called robust stacking. If that's what you mean specifically, you should perhaps make it explicit in the answer. However, I'd remark that robust stacking is actually not much better than simple averaging in case of shot noise; its main advantage is that it prevents extreme outliers from influencing the final result. I.e. it will completely remove an meteor that's passed through the frame at some time. Shot noise isn't extreme enough though, and may in fact end up stronger due to the unequal weights that a robust-stacking assigns. – leftaroundabout Jan 29 '18 at 22:54
8

Digital sensors heat up over long exposures; I've had exposures as short as 6 minutes exhibiting very visible colour distortion in the corners of the frame. Run the chip continually for hours without specialist cooling and the result would likely be an unusable mess.

  • I worked at an observatory once and I can confirm the big telescope's CCD sensors (back in these days) had dedicated cooling systems. I don't remember if it was nitrogen but it was something like that. – Basj Jan 30 '18 at 21:13
  • dedicated deep space astrocams typically have 30-50C of thermometric cooling. Black body radiation is proportional to the temperature (in Kelvin) to the 4th power. So while nowhere near the ~99.5% reduction in thermal noise that LN2 can give they do reduce it by about one third to one half. The lower noise floor is one of the main reasons why people use them instead of conventional DSLRs. – Dan Neely Jan 31 '18 at 1:00
1

Also doing long exposures leave you with the capabilities of the software built into the camera. By taking many images and postprocessing them afterwards you can use software with different capabilities and priorities which may get you a better picture.

1

On reading the other answers here and with the super blood blue moon event being announced the other week I decided to try an experiment. I shot the moon during one of the brief breaks in the cloud when I was not either at work or asleep. I shot handheld as it was just an experiment. My camera, an Olympus E-520 (released 2008), is notoriously noisey and 70-300mm lens of fairly ordinary quality when pushed to its limits, but the four thirds sensor size makes this a 600mm equivalent field of view.

I took several shots at F9.0 1/40s iso800 300mm image stabilisation on.

I selected about 5 and tried using Darktable, Gimp, Siril and Hugin to process the files. The picture shown here has a single colour frame with no noise processing and a composite image made up from 5 similar shots aligned and blended then converted to black and white and sharpened a little. You can see in the corners of the black and white image where the aligned images have not overlaped revealing the effect that additional layers have on noise.

Whether I simply averaged the layers in Gimp (though aligning and rotating the layers was nigh on impossible) or used Hugin or Siril, noise was noticeably reduced in the background.enter image description here

0

A very long exposure is equivalent to averaging multiple shorter exposures. When you take multiple short exposures, you always have the option to do an average to imitate a long exposure if you want. But it also makes many other options (besides averaging) available to you.

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.