Blog · 2026-05-19 · 6 min read

How much time should you spend imaging M31?

M31 (the Andromeda Galaxy) is bright, large, and forgiving — the "first deep-sky target" for many imagers. But what does "done" look like? When should you stop adding sub-frames and move on?

There is no exact answer (but there is a useful range)

Integration time is set by sensor SNR vs sky background, filter, target brightness, focal length and image scale, and how clean a final image you actually want. M31 is bright enough that you can pull a respectable image with one to two hours and still be improving meaningfully past thirty. The right number depends on your taste and your conditions — there isn't one correct value.

The signal-to-noise math, simplified

SNR scales as the square root of integration time. Doubling integration time gives roughly 1.4× the SNR. Going from one hour to four is a meaningful jump; going from thirty hours to thirty-three is barely visible. Diminishing returns set in fast, and they set in faster the more time you've already accumulated.

This is the single most important thing to internalise: the first hour buys you a lot, the tenth hour buys you a bit, the thirtieth hour buys you very little. Past that point, what you do differently matters more than how much more you do.

Practical reference (illustrative — not a prescription)

These ranges are a starting point. Your sky, your sensor, your optics, and your taste will all shift the numbers — sometimes substantially.

Quick image, broadband (L only): 1–3 hours. Enough to see structure, dust lanes, and the bright core.
Solid broadband (LRGB): 8–15 hours total — typically 4–6 hours of L with 1–2 hours each of R, G, B.
Refined broadband: 20–40 hours. Cleaner low-surface-brightness detail in the outer arms, smoother colour, less aggressive noise reduction.
Adding narrowband (HOO for the Ha plumes around M31's halo): 5–15 hours per narrowband filter. M31's faint outer Ha needs the most time of any of its features.
Magazine-cover quality: 60+ hours, usually multi-year accumulations across multiple rigs or sites.

Again: these are illustrative. f-ratio, sky quality (SQM), sensor quantum efficiency, and the look you're going for can each move the numbers up or down by a factor of two.

Why M31 in particular

M31 is unusual in a few ways:

Magnitude 3.4 — by far the brightest galaxy in the northern sky. You collect signal quickly on the core and inner arms.
About three degrees across — usually a mosaic with anything tighter than ~600mm focal length. Plan your framing before you plan your integration budget.
Faint outer Ha emission — the halo arcs and plumes are dim and need disproportionately more integration than the broadband core. The galaxy that looks "easy" hides one of the more demanding narrowband targets in the sky.

When to switch from "more time" to "better time"

Once you've got ten or more hours, the remaining gains come from better sky (lower SQM), better seeing, or different filters — not more of the same. If your image at 10h doesn't show what you want, going to 20h of the same data from the same site probably won't change that. A drive to a darker site, or a swap from broadband to narrowband, often will.

The corollary: if you're three hours in and unhappy, more integration is the right answer. If you're fifteen hours in and unhappy, more integration is usually the wrong answer.

How to know where you are now

All of the above assumes you can answer the question "how many hours do I have on M31, per filter, right now?" Most imagers can't, and not because they're disorganised.

Folder counts mislead — they include calibration frames, test exposures, and mis-labelled files. Sequencer reports only cover what one sequencer captured on one rig during one session, not your archive. The trustworthy answer is to read the FITS headers for every frame in the archive and group by target plus filter. That's exactly what Photon Ledger does: it walks the archive, parses headers, and gives you the per-target per-filter totals so you can compare what you have to what you need. See total integration time by target for the workflow.

Repeat for any target

The same approach — look up a published reference range, compare to what you've actually captured — works for M42, M51, M81, M33, NGC 7000, IC 1396, and anything else you're working on. M31 is just the most common starting point. The signal-to-noise math doesn't care which photons it's counting.