StellacuityBETA

AI sharpening (deconvolution) and noise reduction for your astronomical images. Tighter, rounder stars and cleaner backgrounds - measured from your image's own stars, with nothing else to enter.

Try it free for 7 days Buy - £15

Introductory price of £15, including all future releases of deconvolution and noise reduction.

Version 1.0 beta 3 · Windows 10 / 11 · FITS, XISF and TIFF · mono and colour

See the difference

Drag the slider. Each pair is the same crop of a linear stack, before and after Stellacuity (Sharpen: Normal, then Reduce noise: Normal), shown with the same stretch.

All images taken by John Hawkes, the author of Stellacuity, apart from the final simulated example. Each is a linear master exactly as it came from stacking - no gradient removal and no hand-tuning of any kind; the two colour images have simply been colour calibrated. The only work done on them is Stellacuity's own deconvolution and noise reduction; the stretch is added purely so they can be seen.

Before: the Cocoon Nebula (IC 5146) in colour After Stellacuity: the Cocoon Nebula (IC 5146) in colour
BeforeStellacuity

The Cocoon Nebula (IC 5146), colour: a dense star field tightened, with the nebula's dark lanes sharper and the noise gone.

Before: the Hamburger Galaxy (NGC 3628) in colour After Stellacuity: the Hamburger Galaxy (NGC 3628) in colour
BeforeStellacuity

The Hamburger Galaxy (NGC 3628), colour: a crisper dust lane and halo, tighter stars, a smoother background.

Before: Bode's Galaxy, M81 After Stellacuity: Bode's Galaxy, M81
BeforeStellacuity

M81, Bode's Galaxy: spiral arms and dust lanes sharpen, and the stars tighten.

Before: star clouds and dust lanes in the disk of M31 After Stellacuity: star clouds and dust lanes in the disk of M31
BeforeStellacuity

M31, the Andromeda Galaxy: soft star clouds in the disk break up into individual stars, with the dust lanes sharper.

Before: the Pinwheel Galaxy, M101 After Stellacuity: the Pinwheel Galaxy, M101
BeforeStellacuity

M101, the Pinwheel Galaxy: knots in the spiral arms and the faint outer arms come through more clearly.

Before: Pickering's Triangle in the Veil Nebula After Stellacuity: Pickering's Triangle in the Veil Nebula
BeforeStellacuity

Pickering's Triangle in the Veil Nebula: fine filaments sharpen and the background noise drops away.

Before: the Heart Nebula (IC 1805) After Stellacuity: the Heart Nebula (IC 1805)
BeforeStellacuity

The Heart Nebula (IC 1805): sharper pillars and dark structures, with the star field tightened.

Before: the Pacman Nebula (NGC 281) After Stellacuity: the Pacman Nebula (NGC 281)
BeforeStellacuity

The Pacman Nebula (NGC 281): the dark lanes and globules gain edge detail, and the noise drops away.

Before: the Crab Nebula, M1 After Stellacuity: the Crab Nebula, M1
BeforeStellacuity

M1, the Crab Nebula: the filaments sharpen and the noisy background cleans up.

Before: the Dumbbell Nebula, M27 After Stellacuity: the Dumbbell Nebula, M27
BeforeStellacuity

M27, the Dumbbell Nebula: sharper structure in the nebula and a tighter star field.

Before: the Propeller Galaxy (NGC 7479) After Stellacuity: the Propeller Galaxy (NGC 7479)
BeforeStellacuity

The Propeller Galaxy (NGC 7479): the bar and arms sharpen, and the background noise is cleaned away.

Before: the globular cluster M13 After Stellacuity: the globular cluster M13
BeforeStellacuity

M13, the Great Globular Cluster: more individual stars resolved towards the crowded core.

Before: the Tadpoles in IC 410 After Stellacuity: the Tadpoles in IC 410
BeforeStellacuity

The Tadpoles in IC 410 (H-alpha): sharper tadpole heads and nebula, with the star field tightened.

Before: the corner of a simulated star field with one wave of coma After Stellacuity: the corner of a simulated star field with one wave of coma
BeforeStellacuity

Simulation: the corner of a synthetic star field with one full wave of coma, far worse than a typical telescope. Stellacuity measures the comet-shaped stars from the image itself and restores them round and tight.

What it does

Trained on over 6 million simulated observations, covering more than 1.5 million different star shapes (point spread functions) - built from these effects, alone and in combination:

Atmospheric seeingDefocus & doughnut starsAstigmatismComa Trefoil & higher-order aberrations - coming in future releases Aperture diffractionCentral obstructionColour-dependent focus Tracking driftPeriodic errorWind shakeGuiding kicks Stacking softening
✧

Sharpen (deconvolution)

Tightens stars and recovers fine detail. The star shape is measured across your image from its own stars, so tilted or aberrated corners are handled too. Light, Normal, Heavy or Custom.

◑

Reduce noise

AI noise reduction that leaves stars and faint detail alone - including on a bright sky or nebula background.

▢

Crop

Cut away stacking edges or empty corners first - drag a rectangle, or let Stellacuity find the empty edges.

◉

Mono and colour

FITS, XISF and TIFF, straight from stacking. Colour images are processed channel by channel (experimental).

⚡

Any graphics card

Runs on NVIDIA, AMD or Intel graphics (DirectX 12) - or on the processor, more slowly, if there is none.

⇄

Fits your workflow

Saves 32-bit linear FITS or TIFF, ready for stretching in your usual software. Compare with the original at any time, and undo any step.

Simple, step by step

Load, crop, sharpen, reduce noise, save - with a large zoomable view and "hold to compare" at every step.

The Stellacuity window: the Cocoon Nebula after noise reduction

For developers

Stellacuity can also be driven from your own software. A simple command-line interface exposes the same deconvolution and noise reduction as the window, so imaging suites, scripts and automated pipelines can hand an image to Stellacuity and collect the result - no window, no clicks.

Stellacuity.exe sharpen INPUT OUTPUT [--preset light|normal|heavy] [--sharpening 1.5-8]
                [--halos 0-100] [--nonstellar 1.5-8] [--no-protect] [--cpu]

Stellacuity.exe denoise INPUT OUTPUT [--preset light|normal|heavy] [--strength 0-100] [--cpu]

Files in, files out

Mono or colour FITS, XISF or TIFF in; 32-bit floating-point FITS or TIFF out (chosen by the output's extension), still linear and ready for the next step of your workflow.

Easy to follow

Plain-text progress on standard output - PROGRESS <0-1> <message> and NOTE <message>, ending with DONE <output> or ERROR <message> - with exit code 0 on success, so a progress bar takes a few lines.

Licensing

The command line runs on the user's own copy of Stellacuity and uses its licence, so each user needs a Stellacuity licence. During the free trial it shares the trial's 3 saves.

Planning an integration? We'd be glad to hear from you: support@stellacuity.com

Download

Try Stellacuity free for 7 days on your own images - nothing to sign up for - and save up to 3 results. After that, a licence key unlocks it.

Download Stellacuity 1.0 beta 3 (Windows, 500 MB zip)

Important: right-click the downloaded zip and choose Extract All… first, then run Stellacuity.exe from the extracted folder. Started from inside the zip, it cannot find its own files.

Windows 10 or 11 (64-bit). A graphics card makes it much faster; 8 GB of memory or more recommended.

Install

  1. Right-click the downloaded zip → Extract All… (don't run it from inside the zip).
  2. Open the Stellacuity folder and double-click Stellacuity.exe.
  3. If Windows says "Windows protected your PC", click More info → Run anyway. Stellacuity is not yet signed with a paid certificate, so Windows doesn't recognise it.
  4. Load a linear (unstretched) image and follow the steps.

Buy Stellacuity

£15

Introductory price

One-off payment at the introductory price - buy now and every future release (of deconvolution and noise reduction) is yours at no extra cost, even if the price rises later. Your licence key arrives by email straight away: enter it in Stellacuity (step 1, "Enter licence key") - it is activated once online, then works offline.

Buy a licence - £15

Includes all future releases of deconvolution and noise reduction. Any major new modules added later may be sold separately.

Secure checkout by Polar; VAT handled for you.

Questions

Which images work best?

Linear (unstretched) stacks, before any stretching or other sharpening. For the best results, work in this order:

  1. Crop away the ragged stacking edges.
  2. Remove the gradient, leaving a small positive background rather than clipping it to zero.
  3. Colour calibrate colour images, keeping the data linear.
  4. Sharpen with Stellacuity.
  5. Reduce noise with Stellacuity.
  6. Stretch and carry on with your usual creative processing.
Do I need to enter my telescope, camera or pixel scale?

No. Stellacuity measures the star shape across the image from its own stars.

Does it work with colour (one-shot colour) images?

Yes, experimentally: each colour channel is sharpened and cleaned separately, so it takes about three times as long. Mono images are the main use for now.

Is my image uploaded anywhere?

No. Everything runs on your own computer. Stellacuity only goes online once, to activate your licence key.

How does the trial work?

The full program runs for 7 days from the first time you start it. You can sharpen, reduce noise and compare as much as you like, and save up to 3 finished results. After that it asks for a licence key; your settings and images are untouched.

Why "beta"?

Stellacuity works well but is still being improved - feedback is very welcome, and beta buyers get every update to deconvolution and noise reduction.

About

Stellacuity is made by John Hawkes, an amateur astrophotographer in the United Kingdom.

John studied theoretical physics at the University of Kent in the United Kingdom in the early 1980s, then took a Master of Science in software engineering, specialising in parallel computing. He spent the next twenty years as a software engineer, on work as diverse as computational fluid dynamics, futures-market modelling, operating systems for transputers and a wide range of business applications, before retraining for a career in medicine, where he became a lead MRI radiographer - a working life spent between physics, high-performance software and the art of getting clear images out of difficult data.

Under the night sky, he images from an observatory he built himself in his own garden, where his main instrument is a 12-inch Ritchey-Chrétien telescope and much of the equipment runs on firmware he has written himself. Every example on this page is his own data - the same kind of data you bring home from a night under the stars, with all its soft stars, seeing and corner aberrations.

Stellacuity is the result of a long effort to recover the sharpness that the atmosphere and optics take away, while staying true to the data. It is developed and supported by John himself, so questions, suggestions and examples of your own results are always welcome at support@stellacuity.com.