Ask why someone's image of the same target with the same class of gear looks so much cleaner than yours, and the answer — nine times out of ten — is printed right in their caption: '32 hours'. Integration time is the currency of this hobby. Signal grows linearly with hours while noise grows as the square root, so signal-to-noise improves as √t: four times the hours, twice the quality. Ten hours isn't twice as good as five — it's 41% better. The curve never goes flat, but it does get expensive.
The practical consequences: your first hour is transformative, your tenth is meaningful, your fortieth is a connoisseur's decision. Faint structure — the outer halo of a galaxy, the dust around the Iris, the shockfronts inside a supernova remnant — doesn't 'appear' at some magic threshold; it emerges from the noise floor gradually, and stretching pulls it up only when the noise beneath it is low enough to survive the stretch.
What this changes about how you plan
Think in projects, not sessions. A 20-hour Rosette collected over six nights beats six different 3-hour targets in every way that matters — and it teaches you more, because you calibrate and process one dataset deeply instead of six shallowly.
It also reframes gear lust. The upgrade from 5 to 20 hours on target is free and improves your image by a factor of 2. Almost no $2,000 purchase does that. The gear that DOES matter is the gear that buys you more usable hours: a mount that doesn't throw away subs, automation that keeps shooting after you fall asleep, and planning that puts you on the right target at the right time.
- Set an hours goal per target before you start — 10 h broadband, 20 h narrowband is a decent default — and let the goal, not boredom, decide when you're done.
- Keep shooting a 'done' target on nights that don't suit anything else. The stack only gets better.