Dark Nebulas Detection - Integration time and subexposures

Discussion of using SharpCap for Deep Sky Imaging
Post Reply
AndreVilhena
Posts: 7
Joined: Sun Nov 01, 2020 10:25 am

Dark Nebulas Detection - Integration time and subexposures

#1

Post by AndreVilhena »

Hello everyone,

It is often seen people taking longer subexposures to detect dark structures (eg. M78), tipically at least 5 min. However, reading Dr. Robin Glover detailed explanation in this forum about picking the correct exposure, the best SNR for a given total integration time is achieved with a single subexposure time, which is independent of the object signal, and usually less than 5 min. In addition, the integrated object signal depends on the total integration time, not the subexposure time.

Is the need of longer subexposures to shoot dark structures a myth or is there any theoretical support to it? If the latter, what am I missing?

Cheers,
André
User avatar
turfpit
Posts: 1779
Joined: Mon Feb 13, 2017 8:13 pm
Location: UK
Contact:

Re: Dark Nebulas Detection - Integration time and subexposures

#2

Post by turfpit »

Some real world examples:

https://www.astrobin.com/f9ss9i/?award=top-pick&nc=user, captured using LRGB frames of 300s and Ha of 600s under Bortle 1 skies using top spec equipment ($40,000 16" astrograph & $12,000 camera).

Astrobin Top Picks of M78 https://www.astrobin.com/search/?q=m78& ... =top-pick . Many of these are LRGB and 300s exposure.

This Image of the Day https://www.astrobin.com/wj6sox/?award=iotd&nc=user used a ZWO ASI6200 and 100s exposures (C14 with Hyperstar, makes the system ~f/2).

[EDIT] The above image was also an APOD https://apod.nasa.gov/apod/ap210121.html. Skies were Bortle 3/4.


Theory aside, I keep an open mind and would be interested to see images built using shorter exposures which can match the examples given above.

Dave
User avatar
oopfan
Posts: 1321
Joined: Sat Jul 08, 2017 2:37 pm
Location: New York
Contact:

Re: Dark Nebulas Detection - Integration time and subexposures

#3

Post by oopfan »

Hi Andre,

It has been a while since I last used "The Brain" for exposure determination, so my apologies to Robin if what I am about to say is wrong...

The Brain is primarily concerned with light pollution, not the surface brightness of the object. However, there is a control that allows you to select your "noise tolerance". I recommend selecting "low noise tolerance" if you are interested in capturing dark nebulae or galactic halo. The Brain will recommend a long exposure. However, if you are interested in capturing only the brighter parts, then select "high noise tolerance". The Brain will recommend a short exposure.

Like you, I love dark nebulae. Several months ago, I attempted to capture LDN 1251 from a Bortle 5 site. Learning from previous failures, I realized that I needed a much longer exposure. My calculator said that I needed a 15-minute exposure at bin 1. However, I don't have active guiding so I am limited to a maximum exposure of 90 seconds. At bin 1 and 90-second exposure, it would have taken 30 hours to achieve a reasonable final image. So, I chose to run at bin 3, but I gave up resolution in doing so. Now, the total integration time fell to only 5 hours, something that I could easily achieve in two sessions. Here is a link to that post:

viewtopic.php?f=16&t=3182

Here is another example of using binning to my advantage:

viewtopic.php?f=16&t=3146

To achieve the best image quality, you ought to operate at bin 1, but you will need a looong exposure in order to raise the SNR of the nebula above detection level (i.e. SNR > 1). Of course, I am not suggesting that anyone can capture a dark nebula regardless of light pollution. Long exposures require good guiding and a camera with a deep well to mitigating saturation of bright features like stars. In my case, in the examples I gave, I was mostly concerned with capturing the nebula. All other aspects I was willing to sacrifice.

Also, I must note that my Bortle 5 light pollution consumed 10,000 ADU. If I had Bortle 6 skies, then it could easily have risen to 20,000 ADU or higher. Your camera will saturate the closer you get to a city. This is where a camera with a high full well depth can come in handy. Still, however, I greatly doubt that you can capture a dark nebula at Bortle 9.

Brian
titus67
Posts: 1
Joined: Thu Mar 11, 2021 9:03 am

Re: Dark Nebulas Detection - Integration time and subexposures

#4

Post by titus67 »

We know that the SNR increases with the square root of the number of subexposures. This means that you're going to get diminishing returns in SNR the higher their number goes; at some point you'll reach an asymptote for that length of subexposure. Then how can it be possible to match the SNR of an object shot in magnitude 7 skies versus magnitude 4, for example?

Here's my reasoning, and please tell if I'm wrong. Let's say that the object you're shooting is faint enough that at the pixel level, there is only 1 photon every 5 minutes that reaches the sensor at a particular pixel. If you're shooting 1 minute exposures, that means the photon will only reach the sensor one out of every 5 exposures, on average, but if you're shooting ten minute exposures, it will be in every one. When you average the 1 minute exposures, no matter how many subs you have, the pixel value is only going to be roughly 1/5 of the pixel value in the final stack with the ten minute exposures. Hence, it's impossible to get the same SNR in different limiting magnitude skies, because your subexposure length is limited due to the sky fog. You can't shoot ten minute exposures in magnitude 4 skies, unless you want pretty all-white frames.

Each subexposure has to be long enough so that we do not have to worry about the camera's Read Noise; addressed by my page on Minimal Exposures. Thenceforth, ignore Read Noise. we need to narrow down (better define) the skyfog statistical noise by increasing Integration Time. But the increase in Integration Time will improve the SNR of the skyfog, skyfog signal divided by the quantum-statistical noise, in a predictable fashion.
Post Reply