Software · In active development

AstroDiscovery

From the telescope to the galaxy. Your nights go from subs to finished pictures with no manual steps, then into 3D, and into a flyable map of the real sky.

Every Finished picture, in flight

AstroDiscovery splits each Finished picture into nebula and stars, places every star at its Gaia distance, and records a fly-in. These are real pictures, processed end to end by the app.

The story

Why I built AstroDiscovery

I love the nights. Planning the targets, setting up the rig, watching the live stack slowly pull detail my eyes can’t see out of nothing. That part of astrophotography hooked me straight away.

The processing didn’t. I did the tutorials. I watched the videos. I’d finally get a workflow down pat, step away from the hobby for a few months, and come back having forgotten all of it. Back to the videos. And even when I remembered every step, I wasn’t happy with the results. I didn’t have the hours to split Ha and OIII out of a one-shot-colour camera and learn PixelMath, only to find the data wasn’t good enough in the first place.

Hours of processing can’t save a bad sub. And checking hundreds of subs by hand, one at a time, took forever.

First: sort the subs

Could something sort them for me? Find the darks and flats that match each night’s temperature and settings. Drop the frames where cloud has bloated the stars. And when yet another Starlink satellite cuts across a frame, mask out just that line instead of throwing away the whole sub. That was the start of AstroDiscovery.

Then: process it, end to end

Could the whole thing be scripted? A lone galaxy needs different handling to a target buried in a nebula field. Mosaics need to join without seams. After a lot of hours tuning the pipeline, it started turning out pictures I was genuinely happy with, even from not much data. AI upscaling and noise reduction came next, and honestly? The difference is subtle. The picture before the AI step is already the impressive part.

Then: learn why

Why was one night’s data bad? The Moon was up, it was windy, the seeing was poor. Was 180 seconds really better than 60, once you count the subs lost to wind wobble and satellites? AstroDiscovery started measuring all of it, so the answers come from my own data, not forum rules of thumb.

Then: build it up over time

Once it was measuring everything, it made sense to keep it. AstroDiscovery grew from an imaging tool into a targets-and-fields tool: going back to the same patch of sky over months, combining nights into deeper pictures, and tagging the whole library with every object found in it.

That history opens up the fun bit: watching for change. Variable stars, supernovae in the galaxies I’ve imaged, nebulae that flicker and ripple with light echoes, and new points of light checked against the star catalogues.

The list keeps growing. As the hobby evolves, AstroDiscovery evolves with it, shaped by the discoveries of this great astrophotography community.

How it runs

  1. Your capture store
  2. Home server
  3. Processing computer
  4. Your cloud store

A home server, always on

A small always-on computer such as a Raspberry Pi 5. Collects the subs, analyses every frame, plans, and serves the web app to any browser, on a desktop or a phone.

Mac or PC, for the heavy lifting

Your processing computer picks up jobs whenever it is switched on: Siril, GraXpert, StarNet, Cosmic Clarity and DeepSNR.

Your cloud store, for finished work

Finished pictures go to your cloud store. Old raw nights are archived there too, and brought back when needed.

Chapter 1

The night

You image. By morning every sub is collected, measured and checked.

01

Hands-free capture

Every sub is collected from your capture store, plate-solved and filed while you sleep.

  • Frames are grouped into nights and fields of sky by where they point, so a return visit months later joins the same history.
  • Each field is named for you, even the southern nebulae only catalogued in RCW, Gum or Sharpless.
01 The gallery02 The same on a phone
02

Every frame checked

By morning every sub is measured, satellite trails are found, and the bad frames are flagged.

  • Seeing, transparency, sky brightness, the Moon’s position and brightness, and guiding, for every frame.
  • A score out of 100 for each picture, with its weakest area named and tips from its own data.
01 The frame grid, with “Leave out” on each sub02 The score and how to do better
Chapter 2

Processing

From checked subs to a finished picture, without touching a slider unless you want to.

03

Settings measured, not guessed

What works best for your camera, your sky and your mount, measured from your own subs.

  • Signal-to-noise per hour for each sub length, counting the subs lost to satellites and wind.
  • “Settings to lock in”: sub length, gain and sensor temperature, each with the reason.
01 Every sub measured: banding, read-noise efficiency, star size and sky brightness02 Signal-to-noise per hour for each sub length
04

A professional pipeline

Your processing computer, Mac or PC, runs the full workflow whenever it’s on.

  • Siril, GraXpert, StarNet and DeepSNR, with true star colours from Gaia.
  • Foraxx and HOO palettes for every picture, and seamless mosaics.
01 The processing queue
05

Edit studio, then Finish

Fine-tune live in the browser, then Finish at full size.

  • Finish adds AI noise reduction and a 2× AI upscale.
  • Pictures stay drafts until Finished; only Finished pictures go to your cloud store.
01 The Edit studio on the Carina Nebula
Chapter 3

Explore

Your pictures come alive in 3D, then hang in a flyable map of the real sky.

06

Your pictures in 3D

Every star at its measured Gaia distance. Sway round the picture, or fly into it.

  • Every Finished picture gets a fly-in, ready to share.
  • On a phone, tilt to move the view.
01 M17 in 3D02 The fly-in recorded for every Finished picture
07

Space Explorer

Fly from your own garden to your pictures, hung in space at their real distance.

  • Nearly two million real stars in their true colours, plus the planets, the Milky Way and the Local Group.
  • Starts at tonight’s date and time, with your own horizon.
01 Arriving at the M17 picture02 Above the Milky Way
Chapter 4

Plan and discover

What to shoot next, and what has changed since you last looked.

08

Plan the next night

What to shoot tonight, and how.

  • The dark window, the Moon and a seeing and cloud forecast through the night.
  • Which pictures need more data, and how much tonight’s hours would raise their score.
01 Tonight’s plan
09

Discoveries over time

Your fields, watched for genuine change night after night.

  • New points of light, checked against the catalogues, with a vision model’s second opinion. You decide.
  • Variable stars, changing nebulae and a supernova watch for the galaxies in your fields.
01 Discoveries02 Supernova watch: NGC 300

Status

AstroDiscovery is in active development, and I hope to share the project soon.

Space Explorer imagery: HYG database (CC BY-SA 4.0), ESO/S. Brunier Milky Way (CC BY 4.0), Solar System Scope planet maps (CC BY 4.0), Sentinel-2 cloudless by EOX (CC BY-NC-SA 4.0), Andromeda by Adam Evans (CC BY 2.0).