Mis à jour : janv. 27
One of the question astrophotographers often ask themselves before imaging a target is "how long should my exposures be?". This topic is kind of blurry and not often covered online. In this post, we're going to compare four different stacked images of the Rosette Nebula, all totaling one hour of integration time but with different exposure lengths.
In our previous Galactic Experiment below...
... many of you guys asked us to make a video about comparing several one-hour images stacked using different exposure times, and so here we are!
In the post below, we'll go over four 1-hour stacks using the following frames:
60 x 60 seconds
30 x 120 seconds
12 x 300 seconds
4 x 900 seconds
We will compare each and try to find out which one yields the better result. For this experiment, we took the pictures of the Rosette Nebula with a ZWO ASI071MC camera and the Radian Raptor 61 telescope. We imaged from our Bortle 9 backyard and used the Radian Triad Ultra filter to combat light pollution.
For our comparison, we only stacked the Light Frames and so did not take any flat or dark calibration frames. Hopefully this was the best way to go to have a fair, unbias experiment!
Because we are about to compare unprocessed master files, we will focus on the actual object and not on the surrounding space.
*Note: We used the STF (automatic stretch) for this experiment. It is not the best way to compare the data but we were unable to apply the exact same manual stretch to all. When we tried, one image would be fine but the other three would be either almost completely black or white. Hopefully this does not completely ruin the experiment's results.
Our friend Patrick (check out his awesome Astrobin page!) downloaded our data and helped us compare the results more accurately by scaling each image to the 900sec stack and then matching the mean. You can see his findings and screenshots at the bottom of this post!
60 second exposures
Here is the first image. This is a stack of 60 frames each at 60 seconds. We were immediately pleasantly surprised when opening up this frame on PixInsight. The Rosette Nebula looks great, the cluster within as well, and there seems to be some nice details in the gases around the core.
Overall though, the image lacks contrast and looks a little fade overall. The noise is also pretty noticeable. We can deduct here that the signal received by the sensor over a period of 60 seconds was not enough at all to swamp out the noise.
120 second exposures
Here, we doubled the exposure time but divided in half the number of frames. In the end, the total integration time is still one hour.
When comparing this shot with the previous one, it is noticeably brighter and the amount of noise seems to be just slightly better. There is also more faint gas visible on the outer edges of the object.
For us, 2-minute exposures is a better option than 1-minute exposures here, at least for this particular camera, target, and light pollution level.
If we had to pick between 60 and 120 second exposures based on this comparison, we would go with 120 seconds.
300 second exposures
We then stacked 12 300-second exposures. That is 5 minutes per frame and, once again, we end up with one hour of total integration time.
The main difference that quickly jumped to our eyes here is the color. In this one, the center part of the nebula is fully red, whereas in the two previous attempts, that area had a pale white/blue hue.
This is actually a negative thing because it seems like it would be much more difficult to separate the HA from the OIII in the core of the Rosette during processing.
As for the details and the noise, they are very similar to the previous stack although it does appear like the noise has been greatly reduced.
If we had to pick between 120 and 300 second exposures based on this comparison, we would go with 120 seconds. The noise is better here but the colors in the core are sacrificed. This may be due to the Auto-stretch though.
900 second exposures
Lastly, we took 4 images at 900 seconds each. That is 15 minutes per frame!
Surprisingly, the pale color in the center of the nebula made a comeback! This is great because we'd be able to better process the image without making the entire thing completely red.
There are much better details in this image than in the previous version. The main negative aspect here is the noise. This stack is very noisy, which makes sense since we only stacked four frames.
If we had to pick between 300 and 900 second exposures based on this comparison, we would go with 900 seconds. This is assuming we could spend an entire night imaging the object and end up stacking many frames, which would eliminate most of the noise.
Below is the video we made about comparing these four stacks! We zoom in on each and discuss what we think about them.
From this experiment, it seems like longer exposures give better details, more contrast and great colors, but at the cost of noise. If you are planning on imaging a target for just an hour, an exposure time between 2 and 5 minutes would be ideal. If you plan on imaging an object for an entire night, it seems like choosing a longer exposure time might give you the best results. The noise will easily be taken care of by stacking many frames and, of course, including calibration frames.
Remember that this depends on your camera and the amount of light pollution you are under. Many astrophotographers take some test shots and look at the many numbers that appear in the stats of the image to determine what the best exposure length would be. We are not going to go in depth about it here because honestly, we never cared that much about these numbers in many years of imaging. Today, we usually take 5 to 10 minute exposures for nebulae, 3 minutes for galaxies and 30 seconds for globular clusters, no matter the equipment we use.
A more accurate comparison
As we said in the introduction, our good friend Patrick downloaded our files and was able to more accurately compare them. Using PixInsight, here is what he did to prepare each master:
Scaled the images by the ratio of the exposure time to match intensity
Offset to match the mean
Applied the same stretch to all images
Below you can find a table he made with some juicy numbers (if, unlike us, you like to check out all the stats in your images):
Here is a screenshot showing all four masters. It appears like the 900sec stack is the least noisy one, but also the one that lacks the most crisp details.
Here is Patrick's conclusion after comparing the four masters:
"It is clear that there is less noise in the longer exposure image. One would believe that this is because one incurs a read noise penalty for each stacked frame, however hard to sort out how much might be due to other factors such as varying light pollution/background level.
What is also notable though, it that there is a distinct loss of sharpness in the longer exposure images, and the stars are also larger (larger FWHM) and less round (larger eccentricity). This can result from atmospheric seeing effects and mount tracking errors which blur the stars in a long exposure, whereas those effects can be reduced somewhat with outlier rejection in the stacking process for the shorter exposures. In other words, a sort of "lucky" imaging.
I'm not sure about the origin of the vertical banding in the short exposure stack - I would have thought this would mostly dither out if it were fixed pattern noise.
This would seem to indicate that seeing conditions should also play a role in selecting an exposure length - if seeing is excellent and the mount tracks very well, long exposures will result in less noise for the same total exposure time (the expected result because of the read noise penalty incurred for each frame). If seeing is not good, however, or the mount is not tracking well, a sharper result may be obtained with shorter exposures, however one will need to stack a longer overall integration time in order to achieve the same (lower) level of background noise as the long exposure frame."
A huge thank you to Patrick for helping us analyze this data. We have known Patrick for a couple of years now and his help has been incredible so many times. Give him a follow on Astrobin!
If you're a beginner who would like to learn astrophotography, check out some of our guides below!
And here are our previous Galactic Experiments: