RigTonightStacking and processing astrophotography for beginners: a free, step-by-step workflow
The picture doesn't exist until you process it. A single sub of a nebula looks like a few stars on grey, and even a stacked image looks almost black. Processing pulls the faint signal out of hours of data. It's a skill of its own, and it improves fastest by re-processing the same data set a few times.
This is a beginner workflow using free software. The steps are the same in every program; only the button names change.
The software
- Siril (free, Windows, macOS, Linux): calibration, stacking and full processing, with scripts that do the stacking for you. The best single free choice.
- DeepSkyStacker (free, Windows): simple, reliable stacking. Process the result elsewhere.
- GraXpert (free): removes light-pollution gradients and noise, very well.
- PixInsight (paid, about $300 once): the most powerful, with a steep learning curve. Many people move to it after a year.
- Optional: an image editor (GIMP, Affinity Photo or Photoshop) for final touches.
Step 1: Sort your files
Put each type in its own folder: lights (the target), darks, flats, and bias or dark flats. Throw out lights with clouds, trails or a passing aeroplane right through the middle (or leave the aeroplanes: stacking usually removes them). Siril's scripts expect folders named lights, darks, flats and biases.
Step 2: Calibrate and stack
Calibration subtracts the camera's own signal and divides out dust and vignetting. Registration lines up the stars in every frame. Stacking averages them, rejecting pixels that don't match (satellites, hot pixels, cosmic rays).
In Siril, set the working folder and run the OSC_Preprocessing script for a colour camera or DSLR, or the mono script for a mono camera. In DeepSkyStacker, load each frame type, check all, and stack.
Settings that suit most beginners: average stacking with sigma or winsorised-sigma clipping, and debayering for colour cameras. If you dithered and have plenty of subs, drizzle integration can recover a little resolution from undersampled data.
The result is a linear image: it looks almost black, and that's correct.
Step 3: Crop the edges
Stacking leaves ragged edges where the frames didn't overlap. Crop them off first, because they confuse every later step.
Step 4: Remove the background gradient
Light pollution, moonlight and the horizon leave a brighter glow on one side. Background extraction models it and subtracts it. Use Siril's Background Extraction or GraXpert's automatic mode. Do this while the image is still linear.
Step 5: Fix the colour
For colour data, photometric colour calibration in Siril plate-solves the image and sets the colour balance from the catalogued colours of the stars in it. Then remove any green cast (Siril's Remove Green Noise; cameras with a Bayer filter have twice as many green pixels).
Step 6: Stretch
Stretching brightens the faint parts much more than the bright ones, so the nebula appears without blowing out the stars. Use Siril's Generalised Hyperbolic Stretch or Asinh transform, or an automatic stretch as a starting point.
Go gently. Stretch in a few small steps rather than one big one, and stop before the background turns from dark grey to washed-out grey. It's easy to over-stretch; most beginner images are stretched too hard.
Step 7: Reduce noise
A little noise reduction, applied mostly to the background, goes a long way. GraXpert's denoise and Siril's noise reduction both work well. Too much makes the image look plastic.
Step 8: Stars
Large, bright stars can hide the nebula. Many imagers separate stars from the nebula with a star-removal tool (StarNet, which Siril can run once it's installed), process each separately, then put the stars back smaller and with less glare. Optional, but it's how many of the images you admire are made.
Step 9: Final touches and saving
A gentle increase in saturation, a slight curves adjustment for contrast, and a check that the background is neutral. Save the master in 32-bit FITS or TIFF so you can come back to it, and export a JPEG for sharing.
The most common processing mistakes
- Stretching too far. The background goes grey and noisy. Pull back.
- Skipping flats. Dust donuts and vignetting are very hard to remove later.
- Stretching before removing the gradient. Do background extraction on linear data.
- Too much saturation and noise reduction. The image looks painted.
- Too little data. No processing fixes thirty minutes on a faint target. Collect another night and stack both.
Better data makes easier processing
Most processing problems start at capture. Round stars to the corners need the right back focus; clean data needs cooling or enough dark frames; smooth images need total exposure time and dithering. Check your rig in the rig builder, and plan nights with enough dark hours on one target in the Tonight tab.
New to the whole process? Start with How to get into astrophotography and Your first night of astrophotography.
Questions
What is stacking in astrophotography? Combining many exposures of the same target into one. Random noise averages out while the real signal adds up, so the stacked image is far smoother than any single frame.
How many subs do I need? Think in total time instead: an hour for a bright target, three or more for faint ones. Twenty or more subs also helps the stacking software reject satellites and hot pixels.
Do I need PixInsight? No. Siril and GraXpert, both free, produce excellent images. PixInsight is worth considering once you know what you want it for.