BrettjoAstro · Workflow Part 11 · Shooting Theory
Capturing the Signal
Still not shooting, but learning what it takes. Two families of exposure do the work: the light frames that hold your target, and the calibration frames that strip out what the sensor and the optics quietly add. This covers how many subs, how long, and which calibration frames you actually need.
Light frames and signal to noise
Light frames are exposures of your target, much like ordinary photography, with two differences. They are stored as FITS rather than JPEG or TIFF, which the processing software is built for. And you take a great many of them, not one. The reason is signal to noise.
Signal is the photons arriving steadily from the object. Stars are bright and pour in by the thousand; a faint nebula drops only a handful on the sensor in a given exposure. Noise is everything else: light pollution, a plane or satellite crossing the frame, and the camera's own electronics. In a single exposure the noise can sit almost as strong as the signal.
The trick is that across many exposures the signal stays put while the noise wanders, diffuse and random. Stack the frames and the computer can separate the steady signal from the random noise. More frames means more integration time, a cleaner separation, and less residual noise left behind. That ratio is the whole game.
You can also stop noise at the source: cool the sensor (cooler means quieter), set the ISO or gain to its sweet spot (around ISO 800 on a DSLR; check the documentation for an astro camera's recommended gain), and use a light pollution filter to keep some of it out before it ever lands.
How hard that light pollution filter has to fight is set by where you shoot. The Bortle class and the World Atlas reading from the planning stage decide how much sky glow is competing with your signal in the first place.
Exposure time, the length of a sub
Ten seconds, three minutes, ten minutes? The ceiling comes first from your mount and guiding. A star tracker or alt azimuth mount tops out near ten seconds before field rotation makes a mess. An equatorial mount goes far longer, limited by guiding accuracy against focal length: a 200mm scope might run ten minutes unguided, while a 2000mm scope can blur past one to three minutes even with decent guiding.
Assume perfect guiding and the question becomes how long is wise. The hard limit is blowing out, where bright stars fill the sensor's capacity and go pure white, though modern sensors take a lot before that happens. Some shooters push ten minute subs for faster stacking. The guide argues against it and settles on around three minutes for these reasons:
- Too short loses faint detail
If too few photons from a faint nebula land inside the exposure, the detail is simply not on the sensor, and no processing can recover what was never recorded.
- Too long invites blur
Without perfect guiding, longer subs raise the chance of elongated stars and soft frames.
- Too long costs more when a sub dies
A plane, a gust, a knock to the tripod kills a sub outright. At three minutes you lose three minutes of the night; at ten you lose ten. Going longer buys faster stacking and nothing else, against a bigger risk of lost time.
Three minutes suits nebulae. For a star cluster or galaxy, one minute subs are often plenty. None of this is exact science; it ends up being a feel.
The focal length you matched to the target back in planning is the same number that now sets your guiding tolerance and your safe sub length. Long focal length, tighter guiding, shorter subs.
Integration time, the total
This is the sum of all your subs, and it is hotly debated, with plenty of recent videos offering different ways to calculate it. Two things are settled. A faster scope (lower focal ratio) needs less integration time. And integration time has to climb exponentially to double the signal to noise, so the law of diminishing returns sets in.
In practice, for an average scope around focal ratio four to six: start with five to six hours, process it, and look at it honestly. If it still looks noisy, push toward ten hours and look again.
You can always shoot more. A target is never a closed deal after one night. Shoot it again on the next clear night, add the batches together, go for a third night if you like. Nobody is stopping you.
These are starting points, not rules. The rest is experience and your own standard: shoot many objects at workable quality, which is satisfying early on, or chase top quality from the start. Neither is wrong.
RGB, narrowband and luminance
How the colour is gathered depends on camera and filter:
- RGB
Colour camera with no filter or a simple light pollution filter. On mono, shot through separate R, G and B filters.
- Narrowband
Colour camera with a dual narrowband filter. On mono, dedicated Ha, O3 and S2 filters.
Whenever you shoot narrowband, also grab about twenty short RGB subs, roughly half a minute each, of the same region, and use them for the stars. Narrowband stars come out magenta or greenish and ugly. Real RGB stars cost little time and bring out the best in the image.
You will also hear about luminance, which is essentially shooting with no filter alongside the rest. Ignore it for now; it is a topic of its own.
Dithering
Dithering nudges the scope slightly, up, down, left or right, between every exposure or every second exposure. Because the sensor never sits dead on the same spot, any pattern baked into it is scattered randomly across the frames and resolves away in the stack. Enabling it belongs to the practical shooting in Part 12; drizzle, its relative, comes up in the stacking in Part 13.
Dither or die. You dither no matter which camera you own and whether or not you take dark frames. There is no case where you skip it.
Calibration frames
Calibration frames offset the patterns, effects and imperfections that creep in through the sensor and the optical path. There are four kinds. The clearest explainer on this is Quef the Lazy Geek's recent video, which the guide draws on with his blessing and credit.
Dark frames
ConditionalLight frames with the lens cap on, so they record only the sensor's own contribution. They subtract fixed pattern noise and amp glow, where parts of the sensor read noisier or brighter than others. Modern cameras (ASI 533, 585, 2600 and similar) have no amp glow and need no darks. You need them on a DSLR, an uncooled camera, or an older cooled sensor with amp glow. Check by searching your camera model alongside the term amp glow.
Bias frames
AlwaysLike darks, cap on, but at the fastest possible exposure. They capture the sensor's read noise. Every camera uses bias, whether or not it needs darks. Shoot them once and reuse forever, as long as the camera and temperature stay the same.
Flat frames
EssentialThe bright ones. They do not calibrate the sensor; they correct the optical path, the vignetting at the edges and the dust donuts on filters, sensor, flattener or reducer. Skip them and that shading and those donuts cannot be removed later, and the dust will have moved by morning, so there is no second chance. Shoot at about a third to halfway up the histogram. A flat panel is well worth buying; pointing the scope at sky through a t shirt is hit and miss.
Dark flats
If you shoot flatsWhat darks are to lights, dark flats are to flats: they calibrate the flat frames. They only matter if you are taking flats. Match the exposure length of the flats they pair with, and keep every other setting the same.
N.I.N.A. and ASIAIR both have built in tools for shooting flats.
What you actually need
On a modern cooled astro camera
- Two calibration types only: flats and bias. No darks, no dark flats.
- Bias once: shoot a single master and reuse it for as long as the camera and temperature stay the same.
- Flats every 3 to 6 months if you run an electronic filter wheel, or every session if you swap filters by hand.
- Always dither, regardless of any of the above.
That leaves just two master files, one bias and one flat, to drop into the stack. The stacking itself, along with drizzle, is Part 13.