Do the steps in this order. Every one of them assumes the one before it.
Sort the light frames
Discard anything with trailed stars, a satellite through the middle, or obvious cloud. A frame that is wrong does not average out — it contaminates.
Take and apply the calibration frames
Darks record what the sensor produces on its own; flats record dust and vignetting; bias records the read floor. Subtracting and dividing by them is what makes a stack clean rather than merely averaged.
Register, then stack
Registration aligns the frames on the stars, which matters because the field will have drifted. Stacking then combines them; the method matters less than the calibration did.
Stretch, carefully and repeatedly
The stack is linear: almost everything is crammed into the bottom of the range. Stretching brings it up, and doing it in several gentle steps keeps the stars from bloating.
Stop earlier than you want to
The commonest fault in a first image is over-processing — pushed contrast, crushed blacks, colour that was not there. A restrained version of the same data reads better and is easier to improve later.
The questions that keep arriving
How many dark frames?
Enough that averaging them is meaningful rather than noisy, taken at the same exposure, gain and temperature as the lights. Taking them at the end of the session, with the cap on, costs nothing but time already spent outside.
Why does my stack look grey and flat?
Because it has not been stretched yet. A linear stack is meant to look like that; the faint detail is present and compressed into the bottom of the range.
Last reviewed 17 September 2026