One of the first things we talk about on a Hero Astro workshop is the 500 rule: How the Earth's rotation limits your shutter speeds, requiring us to push the boundaries of our camera's ISO, which results in noisier files than what we would prefer.

One option to overcome this issue is to track the stars. This allows you to expose for far longer at a lower ISO, at the expense of causing problems for landscape shooters, as the landscape itself doesn't move. This technique requires blending the finished tracked star image with a separately shot foreground image. This presents its own challenges, and I personally do not prefer this approach.

There is, though, a halfway house by using image stacking software: Taking a small number of shots and using software to stack them. Apple users have been using Starry Landscape Stacker for some time, leaving us Windows people jealous. But now we have Sequator, a Google hosted project written by Yi-Ruei Wu. Luckily, Google like to keep it simple, and the software is freely available to anyone.

So how does it work? It does a few clever things, like rotating and matching star positions. Additionally, it averages pixels from all the source images to reduce noise. Assuming a shooting situation is limiting a single frame to 30 seconds of exposure, we could feed it four files and have 120 seconds of exposure time. In regards to noise, this simulates shooting at a lower ISO. Instead of ISO 6400, we now have the visual equivalent of ISO 1600, and a cleaner end result. It has a few other features around noise reduction and vignetting, but I will have to go into that another time.

Sequator can read RAW files directly, but I found the results poorer than providing it with pre-processed TIFF files.

Sequator merged directly from RAW files

Sequator merged directly from RAW files

My workflow starts with selecting my four images, processing them in Adobe Camera Raw (or Lightroom), and saving them as TIFF files. I kept the processing pretty standard, so nothing too exciting there (if you want to know more, check out the Hero processing workshops) .

Sony A7R2, 15mm, f2.8, ISO 6400, 30 seconds each.

Sony A7R2, 15mm, f2.8, ISO 6400, 30 seconds each.

I then imported the images into Sequator (File-> Star images). You can set a base image for the foreground if you want, but for this example I didn't bother.


Next, I set an output file name, and I recommend using a TIFF format to retain quality.

In Composition I selected the Align stars & Freeze ground options. For the other options, I found the following worked best for this particular image:


It is worth experimenting with options to see what works best for your stack.

In the next step, I identified the sky by drawing a gradient across the horizon. With a more complex horizon, try the irregular mask.


After clicking the start button, it will show a progress box until it finishes the blending process, and then create the output file.

Once that was done, I loaded the resulting image into Photoshop (or Lightroom) to go through my normal processing steps. Be especially careful with anything that moves (lights on the horizon) and around the horizon boundary, as the blended file may exhibit ghosting effects. If that is the case, either re-create the file with a better mask or fix it with the clone tool in Photoshop.

So, how much improvement will you achieve? Have a look at the following images...

To my eye, the result is better than one stop, but perhaps not quite as good as two stops. It is certainly close, though!

With this particular image, I encountered a downside to using a stacking process. The beams of the aurora, due to their movement, were less distinct. In addition, since you are shooting the same image for far longer, if you combine this with a set of images for your foreground focus, you could be shooting the same composition for five or ten minutes. I still think the extra quality is worth the effort.

Below, the final image has been straightened, levels have been tweaked, and it was run through NIK Colour Efex to create the look. This increases the apparent noise a touch, but the resulting image is still sufficiently low in noise without any additional noise reduction applied. After all, pretty pictures is what we are after!