Drift Stacking Technique

Each year I enter a few images in the Royal Greenwich Museum’s Astrophotographer of the Year competition. This is more out of hope than expectation as the standard is very high. I generally photograph deep sky objects and dabble with lunar, planetary and nightscapes. I figured there’s always someone with a bigger scope, a better camera and endless hours to commit to one deep sky target, so if I wanted a nomination, I’d need to present something they hadn’t seen before. I wondered about trying a technique that I hadn’t seen before combining data from traditional stacked deep sky images with star trailing, typically used on widefield nightscapes. I’m calling it Drift Stacking and I’ll tell you how went about it.

The Rig

For all these examples I used my William Optics WhiteCat. This is the white version of the RedCat (also now available in grey as the SpaceCat). The scope is 250mm focal length at f/4.9, so pretty fast and gives a nice wide view, perfect for large nebulae and wide clusters. The mount is the iOptron Skyguider Pro which is a star tracker (RA axis only) which can also autoguide. I also used a Optolong LPro filter and controlled everything using the ZWO ASIair.

Carbon Stars

I conceived of the idea of combining trailing and stacking in order to capture carbon stars; deep red/ orange supergiant suns. I imagined an image where there was a typical area of sky but with one deep red star standing out against the rest. I knew a few carbon stars like La Superba in CVn. I did a web search for some prominent carbon stars and there were a number in Taurus and Orion that suited my observing window in February 2021.

I selected 119 Tauri as my target and set up my scope one evening. Then I made my first mistake - I found the wrong star! I was using plate solving on the ASIair but because the area was so sparse, it wasn’t finding any prominent objects. I found a deep red star and assumed I’d found my target. In fact, I was a few degrees away on the unexceptional HD38247 - but it was a red supergiant and suited my purposes. It was actually perfect - it was an unremarkable piece of sky with lots of star colour variation.

Sky position & Rotation

Both HD38247 and my second target, M45 The Pleiades were close to the ecliptic. This was coincidental, but probably helped the images as the star trails had very shallow arcs. If I had been looking at targets nearer the zenith, the trails would have been tighter arcs and perhaps not so successful in the composition. I need to try more to see how the object’s altitude affects the Drift images.

I rotated my camera sensor so the image frame long edge aligned (approximately) with the ecliptic. This would mean that the stars should create a pretty flat, straight line as they trail.

Image capture process - Drift

Having found and centred my target I did all my normal camera set up; activated the cooler, found a guide star and got everything working nicely. I took a reference shot; this would be my target frame for the final composition with HD38247 in the dead centre.

To create the Drift images I slewed the RA axis so the camera frame was now completely to the west of the target frame. I wanted to ensure that every star in the target frame would leave a full trail across the whole frame.

I switched off the tracking so the camera was now fixed and started an image run of 60x 60sec exposures. I used SkySafari on my phone to work out how long I would need at 250mm fl to have the target move completely across the static camera sensor. Over the next hour, the target drifted across the fixed camera position from left to right. These subs became the Drift data set.

When I later photographed the Pleiades, I used the exact same process - setting up west, allowing the cluster to drift through, then went back and did a traditional deep sky image run with tracking.

Image capture process - Stack

I then switched on the tracking and slewed west to catch up with my main star, repositioning it in the centre of the frame. I ran off a second set of frames using the star tracking to keep the star centred. I don’t actually recall if I used guiding - I don’t think I did. I used 60sec exposures again, mainly because then I would only need one set of calibration frames. I captured 80-100 frames for the Stack data set.

I then captured calibration frames to be used for both data sets; dark frames x20, bias frames x100 and flat frames x30. For the Pleiades stack I was able to combine a little extra data captured this time last year.

Processing Workflow - APP

Astro Pixel Processor (APP) is my weapon of choice. I loaded in all the Drift Frames and calibration frames and let APP make Master calibration frames of the dark, bias and flats I had taken. Now maybe you’ll laugh at what I did next - I couldn’t find a way to automate this process….I opened each Drift image file with calibration applied in the view options and saved out as a 16-bit .tiff file. I had to do this 60 times for each frame and put them into a folder called Drift. If there’s a way to do this faster please let me know!

I then removed the Drift frames from APP and loaded the Stack frames. I went through the normal APP tabs 0-6 to achieve an integration of my 80-odd 60sec subs. I used the tools on tab 9 to tweak a few bits like light pollution and maybe star colour calibration, then exported the final image as a 16-bit .tiff to a Stack folder.

Processing Workflow - LR

I took the Drift .tiff files into Lightroom (LR) and adjusted all the frames equally; I tweaked the rotation so the trails were perfectly horizontal and increased the saturation a touch. Each frame was a bit noisy as you’d expect, but the next stage would help resolve this. Selecting all the frames, I right clicked and selected ‘open as stack in Photoshop’

Processing Workflow - PS

It took a while to load all the frames across to Photoshop (PS) and it made a very large file - maybe a few hundred GB. As with a normal star trail, I selected all layers and changed the blending mode to ‘Lighten’. This produced an image where each mini-trail was seen alongside the others in a ‘string of pearls’ effect. Because there was a fraction of a second between each frame, it gave me little gaps between each mini trail. Going to each frame one by one, I manually nudged the trail onto the tail of the one before, closing the gaps and making a continuous line.

Leaving this to one side, I opened my Stack .tiff and did some normal PS editing to stretch the data a little, tweak the exposure, contrast and saturation (amongst other little adjustments). Once satisfied with the stack, I did a ‘copy merge’ and pasted this layer into my Draft stack as a separate layer.

For the image of HD38247, the star was dead centre. For the Pleiades, I planned to have the cluster slightly right of centre, so the trailing could enter from the left. The Pleiades image tries to portray energy and movement, so I only wanted some of the drift frames, not a complete left to right passing through the Seven Sisters. So having overlaid the Stack frame, I sorted the layers in the Drift stack into left and right, keeping only those to the left on and hiding the frames to the right.

Smart Object Grouping - PS

Using the Smart Object tool, I converted the left hand frames of the Drift Stack into a Smart Object. This took a few minutes. Once a Smart Object I changed the stack mode to ‘median’. This acts to remove noise and produce a much cleaner image. With my Smart Object of the left hand Drift Frames visible, I did a ‘copy merge’ and pasted into a new file. On a second layer I pasted in my Stack frame. I now had a image that resembled the final composition and I could tweak the relative prominence of either the Stack or the Drift. At this stage you could choose ditch the original PS file if you don’t have GBs of memory to spare.

The combined Drift Frame was nice and smooth, but did have hot pixels visible, so I had to manually pan across and remove these with the clone tool. Then I made final adjustments for brightness and contrast.

The final image of HD38247 combined a full set of Drift frames over a Stack frame. The idea here was to accentuate the star colours allowing the star that produced the colour to ‘sit’ on the line it created. It showed a number of other red /orange supergiants in the frame and given it was the wrong target, I was pleased with the result.

The Pleiades image was trying to present a familiar target in a new way, also drawing out the range of coloured stars around the cluster, which is famous for its young hot blue/white stars. For the submission I called this M45 WHAMM! as it had a certain Pop Art quality and energy about it.

Alternative take

The third image I shared on Twitter was using the same M45 data. I sense most people didn’t spot that this image is The Pleiades too, and is composed of just one 60sec drift sub, replicated 6 times (Seven Sisters yeah?) to make each individual star into a ‘block’. The purpose of this was the star colours were easier to distinguish when ‘beefed up’ and the colour variation is pretty remarkable. I also love the way they sit atop of one another and give an the image depth and an almost 3D-like quality.

In PS I made 6 duplicates of the selected frame, and nudged each one up so they sat one on top of one another with a narrow spacing. I very subtly reduced the opacity of the stripes so the top most line is 100%, dropping by 10% opacity with each stripe. I could still see the familiar Pleaides shape in the image and it also reminded me of some 1980s vinyl album covers I gazed at as a child.

Drift Stack Gallery

I hope you enjoyed this explanation of the process I used and perhaps you’re inspired to have a go yourself? With so much data you can find multiple ways to assemble, crop and rotate into a final image like these examples below.

My images didn’t get a competition nomination, but the process was rewarding and I enjoyed setting myself a challenge and trying something new. Even though they ended up on the ‘No’ pile, I hope the judges paused over the delete button for a few seconds longer than usual. The reaction on Twitter was very positive, so I’m taking that as a win.