Sign in

Guide

How to process Seestar FITS files

A Seestar stack is linear data. Opened as it is, it looks almost black with a few stars, because nearly all of the signal sits just above the noise. Processing turns it into a photo. There are three ways to do that, from the least work to the most control.

All three start from the FITS file, not the JPG. If you don’t know where it is, read Where your Seestar or Dwarf 3 saves its FITS files first.

1. The telescope app’s own image

The Seestar app already stretches the stack and shows you the result: the JPG in the album. The Dwarf app does the same. For a look on the phone that is often enough.

The Seestar app also has AI Enhance, built in, for deep-sky and Milky Way photos. Its settings include noise reduction, star reduction and 2×2 binning, it can run while the telescope is still stacking, and it saves the result to your phone. It is the fastest way to a nicer picture. What it does is decided by the app: you choose among its settings, not how each step works.

2. By hand, in Siril and GraXpert

Siril and GraXpert are free, open-source programs for Windows, macOS and Linux, and Siril has a tutorial written for Seestar files. GraXpert can run on its own or from inside Siril. The usual order:

  1. Stack, if you kept the raw subs. Siril’s Seestar_Preprocessing script registers and stacks the files from the _sub folder. Since Siril 1.4 you fetch it from the Scripts tab in the preferences. A Stacked_….fit skips this step.
  2. Remove the background gradient. Light pollution, the moon and the horizon put a slow ramp of brightness across the frame. Siril’s background extraction fits it from sample points you place on empty sky; GraXpert’s does it with a trained model.
  3. Calibrate the colour. Plate solve the image (the Seestar writes the coordinates into the FITS header, so this is usually one click), then run photometric colour calibration. It matches your stars to a catalogue and sets the colour balance from them.
  4. Reduce the noise. GraXpert’s denoise works best on the linear data, before the stretch, while noise still looks like noise.
  5. Stretch. A histogram or hyperbolic stretch lifts the faint signal into view. This is the step that decides the look of the photo, and the one worth the most practice.
  6. Finish. Saturation, perhaps star reduction, a crop, then save as 16-bit TIFF or PNG and a JPG to share.

This gives you control over every step, and a patient person will get the best image this way. It also takes time: time to learn the tools, and time for every image. PixInsight follows the same steps; it is paid software, used by many experienced imagers.

3. Upload it to Stelora

Stelora runs the same kind of steps for you, with Siril and GraXpert doing the heavy lifting. You upload the FITS file in your browser, from a phone or a computer, and get the photo back, usually a few minutes later. In order, the pipeline:

  • stacks raw subs in Siril, if you sent them, and crops the stacking edges;
  • removes the background gradient with GraXpert;
  • removes noise with GraXpert, on the linear data;
  • sets the white balance from the colour of the stars in your frame;
  • stretches with one curve that adapts to the frame;
  • adds a little saturation, corrects star colour and lifts local contrast.

You get a 16-bit PNG at full size and a JPG, with no watermark. How it works describes every stage, and a finished sample shows a real result you can open now.

What it does worse than a skilled person, as How it works says too:

  • The stretch is one global curve. A hand-tuned stretch in Siril or PixInsight beats it, and probably always will.
  • The noise reduction has one fixed strength. It cleans the sky and takes some very faint texture with it.
  • The background model cannot always tell faint outer nebulosity from gradient, so the faintest edge of a large object can lose a little.
  • Colour corrections are capped. A strong colour cast is reduced, not removed, because an automatic correction that misreads a frame does more harm than the cast.

Which one to choose

For a quick picture on the phone, the app’s image or AI Enhance is enough. If you enjoy processing and want every decision to be yours, learn Siril and GraXpert; nothing automatic will match a good result by hand. If you want a clean, full-resolution photo of tonight’s capture without installing anything, upload it to Stelora. And if you already process well yourself, keep doing that.

Upload your FITS file Open a finished sample