Image stacking - astro-style........

jerry12953

Suspended / Banned
Messages
12,421
Name
Jeremy Moore
Edit My Images
No
I'm trying to find a simple explanation of this technique which seems to be widely used in astrophotography. But all I can find is written in such scientific terminology that I can't understand it.

Can anyone help? Thanks.

"Image stacking is the technique of merging multiple images of the same object, and processing it in a way that increases resolution, decreases noise and artifacts, and multiplies the brightness of any single image. What this means in astrophotography is that, instead of taking one enormously long exposure (which will be susceptible to noise from the camera as well as resolution and trailing issues), you can take multiple small exposures and then stack them on top of each other to produce an image that has good brightness, contrast, and resolution.".

Is it possible to do this in Affinity / Photoshop or even Lightroom?
 
Yes I've done it many times.

There is dedicated software for it (free on windows, paid for on mac) so I've only used photoshop.

Incidentally this tenchnique and its advantages could be used in other types of photography. Long exposures with high dynamic range allowing smooth water/clouds and ultra clean shadows to retain highlights and push shadows...

Regarding Astro from an Astro landscape point of view primarily not deep sky stuff. First, you need to decide if you want to do star trails or static objects, moon, Milky Way etc.

Star trails first, rather than shooting one long exposure which could easily be ruined by a torch pointed at your camera, and cloning satellites, planes out is harder than the stacked method where you conceal them on a layer mask. Also thermal noise from the camera heating up can be a problem.

Think of the main compromise of a long exposure, you are actually recording 2 exposures. One for the static base image, foreground/sky and one for the 'moving' parts of the image, the stars. Well the stars are brighter obviously, but are only in one position on the sensor for a certain amount of time. So the only thing that effects the brightness of the star trail part of the exposure is ISO and aperture. Exposure time does not affect exposure level of the stars, only the length of the trails. But all 3 parts of the exposure triangle do affect the base image providing there is some (however small) illumination. Now what are the chances that firstly the settings you want for the stars will also suit the foreground/sky and, the optimum exposure if the foreground will arrive at the same time you require for the length of the trails?

So, what you do instead is shoot your base exposure for the foreground at a relatively low ISO, stopped down aperture if you require correctly focussed and a long exposure. This can be done during the blue hour or by moonlight to give you the ambient light levels to illuminate the scene. Then at full dark or after moon set you refocus on the brightest star, and shoot with a higher ISO (most cameras) wider aperture and shorter exposures. Probably 10-30 seconds depending on focal length. You are looking to capture relatively sharp stars, and lots of them, shoot with an intervalometer and take hundreds of shorter continuous exposures. In post you process your base image, then separately stack your star images with a lighten stack mode to reveal the lighter pixels on each one (the moving stars) what I like to do is fade the opacity to give tail fade effect. You can then use layer masks to remove unwanted objects. Then flatten down and bring in on top of your base image, either with lighten mode if you want a brighter sky, or as a mask to ground if you want the darker sky from the flattened star image.

Now, more in answer to your question using image averaging to improve iq in Astro when not star trails, or as I mentioned other photography. You know the way cameras produce noise, and are probably familiar with pattern noise at lower ISO (canon more prevalent) caused by the sensor electronics. You will also know the way photons are collected naturally creates noise, this I think is called shot noise. What some Astro guys do is increase the ISO setting until the random shot noise overwhelms the fixed pattern read noise of the sensor. What we are doing with averaging is taking this a stage further, the random shot noise moves around constantly, so from one exposure to the next it will not be in the same place on the sensor. So, by aligning the stars in say up to a dozen images then using an average blend mode (mean or mode I think) photoshop will look at them and things that are the same stay, and things that differ (the random moving noise) are removed. Voila, your image is cleaner, and slightly sharper/brighter because stars twinkle due to upper atmosphere seeing so you get the best stars too hopefully.

When I'm at my computer I'll see if I can post some examples up.
 
Yes I've done it many times.

There is dedicated software for it (free on windows, paid for on mac) so I've only used photoshop.

Incidentally this tenchnique and its advantages could be used in other types of photography. Long exposures with high dynamic range allowing smooth water/clouds and ultra clean shadows to retain highlights and push shadows...

Regarding Astro from an Astro landscape point of view primarily not deep sky stuff. First, you need to decide if you want to do star trails or static objects, moon, Milky Way etc.

Star trails first, rather than shooting one long exposure which could easily be ruined by a torch pointed at your camera, and cloning satellites, planes out is harder than the stacked method where you conceal them on a layer mask. Also thermal noise from the camera heating up can be a problem.

Think of the main compromise of a long exposure, you are actually recording 2 exposures. One for the static base image, foreground/sky and one for the 'moving' parts of the image, the stars. Well the stars are brighter obviously, but are only in one position on the sensor for a certain amount of time. So the only thing that effects the brightness of the star trail part of the exposure is ISO and aperture. Exposure time does not affect exposure level of the stars, only the length of the trails. But all 3 parts of the exposure triangle do affect the base image providing there is some (however small) illumination. Now what are the chances that firstly the settings you want for the stars will also suit the foreground/sky and, the optimum exposure if the foreground will arrive at the same time you require for the length of the trails?

So, what you do instead is shoot your base exposure for the foreground at a relatively low ISO, stopped down aperture if you require correctly focussed and a long exposure. This can be done during the blue hour or by moonlight to give you the ambient light levels to illuminate the scene. Then at full dark or after moon set you refocus on the brightest star, and shoot with a higher ISO (most cameras) wider aperture and shorter exposures. Probably 10-30 seconds depending on focal length. You are looking to capture relatively sharp stars, and lots of them, shoot with an intervalometer and take hundreds of shorter continuous exposures. In post you process your base image, then separately stack your star images with a lighten stack mode to reveal the lighter pixels on each one (the moving stars) what I like to do is fade the opacity to give tail fade effect. You can then use layer masks to remove unwanted objects. Then flatten down and bring in on top of your base image, either with lighten mode if you want a brighter sky, or as a mask to ground if you want the darker sky from the flattened star image.

Now, more in answer to your question using image averaging to improve iq in Astro when not star trails, or as I mentioned other photography. You know the way cameras produce noise, and are probably familiar with pattern noise at lower ISO (canon more prevalent) caused by the sensor electronics. You will also know the way photons are collected naturally creates noise, this I think is called shot noise. What some Astro guys do is increase the ISO setting until the random shot noise overwhelms the fixed pattern read noise of the sensor. What we are doing with averaging is taking this a stage further, the random shot noise moves around constantly, so from one exposure to the next it will not be in the same place on the sensor. So, by aligning the stars in say up to a dozen images then using an average blend mode (mean or mode I think) photoshop will look at them and things that are the same stay, and things that differ (the random moving noise) are removed. Voila, your image is cleaner, and slightly sharper/brighter because stars twinkle due to upper atmosphere seeing so you get the best stars too hopefully.

When I'm at my computer I'll see if I can post some examples up.


Craig,

That's a fantastically detailed response so thanks very much. I was wondering if something like that could be used in a different situation. I'm thinking of going on a Northern Lights Cruise, which is basically a service vessel running up and down the Norwegian Coast right up into the arctic and back. It doesn't stop anywhere for very long so most of the time would be at sea. To do any aurora photography I'd be battling against two main problems

1) ship movement, and
2) engine vibration.

My guess is that a tripod would not help due to (2) in particular, so instead of doing hand-held long exposures with both (1) and (2), would it be possible to take a series of 20 exposures at, say, 1/10th sec hand-held and combine them in software to give the equivalent of a 2 sec exposure? What do you think?

I don't have PS but if it's possible in PS, it must also be possible in Affinity Photo, and I'd be happy to get into that.
 

No, that's still really useful. I've listened to the sequence he uses in LR and he explains it really well.

I'll look at the PS bit later........

Ah, I don't think that was what I was looking for, as you need a series of correctly exposed images to start with. I was hoping to combine a series of underexposed images to get a correctly exposed one.........
 
Last edited:
Images can be stacked in Affinity Photo, I have done it a number of times.
 
Jeremy, if you want any further information, send me a PM.
 
Craig,

That's a fantastically detailed response so thanks very much. I was wondering if something like that could be used in a different situation. I'm thinking of going on a Northern Lights Cruise, which is basically a service vessel running up and down the Norwegian Coast right up into the arctic and back. It doesn't stop anywhere for very long so most of the time would be at sea. To do any aurora photography I'd be battling against two main problems

1) ship movement, and
2) engine vibration.

My guess is that a tripod would not help due to (2) in particular, so instead of doing hand-held long exposures with both (1) and (2), would it be possible to take a series of 20 exposures at, say, 1/10th sec hand-held and combine them in software to give the equivalent of a 2 sec exposure? What do you think?

I don't have PS but if it's possible in PS, it must also be possible in Affinity Photo, and I'd be happy to get into that.

A number of problems at work here!

1. Earth's rotation giving the appearance of stars moving.

2. Camera position moving relative to foreground.

3. Camera not completely stable due to engine vibration.

I like the idea of shorter exposures, then stacked, the problem is you need an anchor point/s. (No pun intended!) If you shoot with foreground, some land on the horizon included, then just align this in each shot. That will allow for camera position movement, (but not camera shake or motion blur) it won't prevent star trailing but shooting wide for a total accumulated exposure of less than say 20-30 seconds you should be fine anyway. I'd still shoot off a tripod though, with a remote release in continuous, (or intervalometer).

To decide how long those exposures should be I think you will actually need to do some test shots on the ship. I would actually do that then make my decision on how to frame up. Take a test shot pointing at something on the ship, try a sequence of test shots, even up to longer exposures. You would be amazed, due to effectively in camera averaging sometimes a small vibration part way through a long exposure is not visible, because the parts of the exposure during the time the camera is not shaking burn onto the sensor for more of the time and you don't see the small shake. Once I'd established what the limit for ship vibration would allow I'd hope to take some auroras when the ship was not sailing, but if it was going to be, hopefully on calm water, I'd during the day (do some test shots framed up on some distant land) to see how quickly you get motion blur on the land. This then gives you a maximum exposure length from the camera's relative movement to the land point of view.

Pretend you get to 1 second maximum, that would be ideal as it could be usable. Remember what I said earlier about the iso and aperture being the only parts of the exposure triangle that are relevant to the exposure of bright moving things like stars. Well that applies to stars after the 500 rule, basically beyond the time that they have moved pixel position on the sensor and are being recorded on a different pixel. Well, with auroras I guess that would happen sooner, as they appear to move quicker, and in post, we are just going to add the green blur together! Pick an ISO and aperture that brings the aurora out as bright as possible, say ISO1600 and f4 minimum. Don't worry if it appears bright, or dark even, the real magic will happen in post with the additive of all the exposures, but with more control over dynamic range headroom.

I still think you will struggle, and what ISO you need to go to will be affected by the length of the exposure you can achieve. If you have a faster lens, then use it. My only other thoughts are to frame up without the land, but include some bright stars, then use these as anchor points to re-align the images in PS. You will need to ensure that the images are sharp though, and not ruined by ship movement still, but my head is hurting from thinking if the stars are corrected for earths rotation what happens to the aurora movement?!


Tail Fade Star Trails;
Insignificant by Craig Hollis, on Flickr

Blue Hour Blend with Image Averaged Milky Way;
Crowfield Milky Way by Craig Hollis, on Flickr

Aurora, f/2.0 20 seconds ISO 800
Jokulsarlon Aurora Borealis by Craig Hollis, on Flickr
 
Thanks all for the suggestions. I'm going to have to give this all a lot of thought; but I've already learned a lot. I do still have three months to go (assuming there are still places available........).

Craig, Those are great images. you obviously have this technique down to a T. I remember one of yours of a waterfall in the Brecon Beacons; you had managed to get a circular eddy of foam below the falls to perfection. I was very impressed with that! I may come back with some questions when i have mulled this over.
 
Thanks all for the suggestions. I'm going to have to give this all a lot of thought; but I've already learned a lot. I do still have three months to go (assuming there are still places available........).

Craig, Those are great images. you obviously have this technique down to a T. I remember one of yours of a waterfall in the Brecon Beacons; you had managed to get a circular eddy of foam below the falls to perfection. I was very impressed with that! I may come back with some questions when i have mulled this over.

I meant to add the frustration of a cruise from a landscape photography point of view is why we have never pulled the trigger on one.

My wife says, "oh but you can get off at these pretty places and take your camera with you", but doesn't appreciate, and nor would the rest of the passengers thatI'd like to be off an hour before sunrise till an hour after, then an hour before sunset and into the night if conditions permitted. What they actually do is an middle of the day excursion...

Your too kind, I've got lots to practice and improve. Incidentally the waterfall picture you mention does the opposite of averaging, so I was taking the additional things, i.e. the moved lead swirl and adding them in.

An understanding of what is capable in photoshop drives the way I use the camera in the field. It is this that I try and teach people, the more you know about developing your shots the better job you can do of taking them. Or, the more problems you can think around...
 
Back
Top