How to fail at astrophotography

This is part 1 of what I hope will become a series of posts. I’m going to focus in this post on my getting started and some mistakes I made on the way.

So, back in 2017 I got a telescope. I fancied trying to do some astrophotography – I saw people getting great results without a lot of kit, and realised I could dip my toe in too. I live between a few towns, so get “class 4” skies – meaning that I could happily image a great many targets from home. I’ve spent plenty of time out at night just looking up, especially on a moonless night; the milky way is a clear band, and plenty of eyeball-visible targets look splendid.

So I did some research, and concluded that:

  • Astrophotography has the potential to be done cheaply but some bits do demand some investment
  • Wide-field is cheapest to do, since a telescope isn’t needed; planetary is way cheaper than deep-sky (depending on the planet) to kit out for, but to get really good planetary images is hard
  • Good telescopes are seriously expensive, but pretty good telescopes are accessibly cheap, and produce pretty good results
  • Newtonians (Dobsonians, for visual) give the absolute best aperture-to-cash return
  • Having a good mount that can track accurately is absolutely key
  • You can spend a hell of a lot of cash on this hobby if you’re not careful, and spending too little is the fastest path there…

So, having done my research, the then-quite-new Skywatcher EQ6-R Pro was the obvious winner for the mount. At about £1,800 it isn’t cheap, but it’s very affordable compared to some other amateur-targeted mounts (the Paramount ME will set you back £13,000, for instance) and provides comparable performance for a reasonable amount of payload – about 15kg without breaking a sweat. Mounts are all about mechanical precision and accuracy; drive electronics factor into it, of course, but much of the error in a mount comes from the gears. More expensive mounts use encoders and clever drive mechanisms to mitigate this, but the EQ6-R Pro settles for having a fairly high quality belt drive system and leaves it at that.

Already, as I write this, the more scientific reader will be asking “hang on, how are you measuring that, or comparing like-for-like?”. This is a common problem in the amateur astrophotography scene with various bits of equipment. Measurement of precision mechanics and optics often requires expensive equipment in and of itself. Take a telescope’s mirror – to measure the flatness of the surface and accuracy of the curvature requires an interferometer. Even the cheap ones cooked up by the make-your-own-telescope communities take a lot of expensive parts and require a lot of optics know-how. Measuring a mount’s movement accurately requires really accurate encoders or other ways to measure movement very precisely – again, expensive bits, etc. The net result of this is that it’s very rare that individual amateurs do quantitative evaluation of equipment – usually, you have to compare spec sheets and call it a day. The rest of the analysis comes down to forums and hearsay.

As an engineer tinkering with fibre optics on a regular basis, spec sheets are great when everyone agrees on the test methodology for the number. There’s a defined standard for how you measure insertion loss of a bare fibre, another for the mode field diameter, and so on. A whole host of different measurements in astrophotography products are done in a very ad-hoc fashion, vary between products and vendors, and so on. Sometimes the best analysis and comparison is being done by enthusiasts that get kit sent to them by vendors to compare! And so, most purchasing decisions involve an awful lot of lurking on forums.

The other problem is knowing what to look for in your comparison. Sites that sell telescopes and other bits are very good at glossing over the full complexity of an imaging system, and assume you sort of know what you’re doing. Does pixel size matter? How about quantum efficiency? Resolution? The answer is always “maybe, depends what you’re doing…”.

Jupiter; the great red spot is just about visible. If you really squint you can see a few pixels that are, I swear, moons.

This photo is one of the first I took. I had bought, with the mount, a Skywatcher 200PDS Newtonian reflector – a 200mm or 8″ aperture telescope with a dual-speed focuser and a focal length of 1000mm. The scope has an f-ratio of 5, making it a fairly “fast” scope. Fast generally translates to forgiving – lots of light means your camera can be worse. Visual use with the scope was great, and I enjoyed slewing around and looking at various objects. My copy of Turn Left at Orion got a fair bit of use. I was feeling pretty great about this whole astrophotography lark, although my images were low-res and fuzzy; I’d bought the cheapest camera I could, near enough, a ZWO ASI120MC one-shot-colour camera.

Working out what questions to ask

The first realisation that I hadn’t quite “gotten” what I needed to be thinking about came when I tried to take a photo of our nearest galaxy and was reminded that my field of view was, in fact, quite narrow. All I could get was a blurry view of the core. Long focal length, small pixel sizes, and other factors conspired to give me a tiny sliver of the sky on my computer screen.

M31 Andromeda; repaired a bit in PixInsight from my original, still kinda terrible

Not quite the classic galaxy snapshot I’d expected. And then I went and actually worked out how big Andromeda is – and it’s huge in the sky. Bigger than the moon, by quite a bit. Knowing how narrow a view of the moon I got with my scope, I considered other targets and my equipment. Clearly my camera’s tiny sensor wasn’t helping, but fixing that would be expensive. Many other targets were much dimmer, requiring long exposures – very long, given my sensor’s poor efficiency, longer than I thought I would get away with. I tried a few others, usually failing, but sometimes getting a glimmer of what could be if I could crack this…

Raw stack from an evening of longer-exposure imaging of NGC891; the noise is the sensor error. I hadn’t quite cracked image processing at this point.

It was fairly clear the camera would need an upgrade for deep space object imaging, and that particular avenue of astrophotography most appealed to me. It was also clear I had no idea what I was doing. I started reading more and more – diving into forums like Stargazer’s Lounge (in the UK) and Cloudy Nights (a broader view) and digesting threads on telescope construction, imaging sensor analysis, and processing.

My next break came from a family friend; when my father was visiting to catch up, the topic of cameras came up. My dad swears by big chunky Nikon DSLRs, and his Nikon D1x is still in active use, despite knackered batteries. This friend happened to have an old D1x, and spare batteries, no longer in use, and kindly donated the lot. With a cheap AC power adapter and F-mount adapter, I suddenly had a high resolution camera I could attach to the scope, albeit with a nearly 20-year-old sensor.

M31/M110 Andromeda, wider field shot, Nikon D1x – first light, processed with DeepSkyStacker and StarTools

Suddenly, with a bigger sensor and field of view, more pixels (nearly six megapixels) I felt I could see what I was doing – and suddenly saw a whole host of problems. The D1x was by no means perfect; it demanded long exposures at high gains to get anything, and fixed pattern noise made processing immensely challenging.

M33 Triangulum, D1x, processed with DeepSkyStacker and PixInsight

I’d previously used a host of free software to “stack” the dozens or hundreds of images I took into a single frame, and then process it. Back in 2018 I bought a copy of StarTools, which allowed me to produce some far better images but left me wanting more control over the process. And so I bit the bullet and spent £200 on PixInsight, widely regarded as being the absolute best image processing tool for astronomical imagery; aside from various Windows-specific stability issues (Linux is rock solid, happily) it’s lived up to the hype. And the hype of its learning curve/cliff – it’s one of the few software packages for which I have purchased a reference book!

Stepping on up to mono

And of course, I could never fully calibrate out the D1x’s pattern noise, nor magically improve the sensor quality. At this point I had a tantalisingly close-to-satisfying system – everything was working great. My Christmas present from family was a guidescope, where I reused the ASI120MC camera, and really long exposures were starting to be feasible. And so I took a bit of money I’d saved up, and bit the hefty bullet of buying a proper astrophotography camera for deep space observation.

By this point I had a bit of clue, and had an idea of how to figure out what it was I needed and what I might do in the future, so this was the first purchase I made that involved a few spreadsheets and some data-based decisions. But I’m not one for half-arsing solutions, which became problematic shortly thereafter.

The scope and guidescope, preparing for an evening of imaging on a rare weekend clear night
M33 Triangulum; first light with the ASI183MM-PRO. A weird light leak artefact can be seen clearly in the middle of the image, near the top of the frame

Of course, this camera introduces more complexity. Normal cameras have a Bayer matrix, meaning that pixels are assigned a colour and interpolation fills in the colour for adjacent pixels. For astrophotography, you don’t always want to image red, green or blue – you might want a narrowband view of the world, for instance, and you for various reasons want to avoid interpolation in processing and capture. So we introduce a monochrome sensor, add a filter wheel in front (electronic, for software control), and filters. The costs add up.

The current finished imaging train – Baader MPCC coma corrector, Baader VariLock T2 spacer, ZWO mini filter wheel, ASI183MM-PRO

But suddenly my images are clear enough to show the problems in the telescope. There’s optical coma in my system, not a surprise; a coma corrector is added to flatten the light reaching the filters and sensor.

I realise – by spending an evening failing to achieve focus – that backfocus is a thing, and that my coma corrector is too close to my sensor; a variable spacer gets added, and carefully measured out with some calipers.

I realise that my telescope tube is letting light in at the back – something I’d not seen before, either through luck or noise – so I get a cover laser cut to fix that.

It turns out focusing is really quite difficult to achieve accurately with my new setup and may need adjusting between filters, so I buy a cheap DC focus motor – the focuser comes to bits, I spend an evening improving the tolerances on all the contact surfaces, amending the bracket supplied with the motor, and put it back together.

To mitigate light bouncing around the focuser I dismantled the whole telescope tube and flock the interior of the scope with anti-reflective material, and add a dew shield. Amongst all this, new DC power cables and connectors were made up, an increasing pile of USB cables/hubs to and from the scope added, a new (commercial) software package added to control it all, and various other little expenses along the way – bottles of high-purity distilled water to clean mirrors, and so on.

Once you’ve got some better software in place for automating capture sessions, being able to automatically drive everything becomes more and more attractive. I had fortunately bought most of the bits to do this in dribs and drabs in the last year, so this was mostly a matter of setup and configuration.

It’s a slippery slope, all this. I think I’ve stopped on this iteration – the next step is a different telescope – but I’ve learned a hell of a lot in doing it. My budget expanded a fair bit from the initial purchase, but was manageable, and I have a working system that produces consistently useful results when clouds permit. I’ve got a lot to learn, still, about the best way to use it and what I can do with it; I also have a lot of learning to do when it comes to PixInsight and my image processing (thankfully not something I need clear skies for).

… okay, maybe I’d still like to get a proper flat field generator, but the “t-shirt at dusk” method works pretty well and only cost £10 for a white t-shirt

Settling in to new digs

Now, of course, I have a set of parts that has brought my output quality significantly up. The images I’m capturing are good enough that I’m happy sharing them widely, and even feel proud of some. I’ve even gotten some quality-of-life improvements out all this work – my evenings are mostly spent indoors, working the scope by remote control.

Astrophotography is a wonderful collision of precision engineering, optics, astronomy, and art. And I think that’s why getting “into” it and building a system is so hard – because there’s no right answer. I started writing this post as a “all the things I wish someone had told me to do” post, but really when I’m making decisions about things like the ideal pixel size of my camera I’m taking an artistic decision that is underpinned by science and engineering and maths – it has an impact on what pictures I can take, what they’ll look like, and so on.

M33 Triangulum, showing clearly now the various small nebulas and colourful objects around the main galaxy. The first image I was genuinely gleeful to produce and share as widely as I could.
The Heart Nebula, not quite centred up; the detail in the nebulosity, even with this wideband image, is helped tremendously by the pixel oversampling I achieve with my setup (0.5 arcseconds per pixel)

But there’s still value in knowing what to think about when you’re thinking about doing this stuff. This isn’t a right answer; it’s one answer. At some point I will undoubtedly go get a different telescope – not because it’s a better solution, but because it’s a different way to look at things and capture them.

So I will continue to blog about this – not least because sharing my thoughts on it is something I enjoy and it isn’t fair to continuously inflict it on my partner, patient as she is with my obsession – in the hopes that some other beginners might find it a useful journey to follow along.

Leave a Reply