Astrophotography Image Stacking, Explained Simply

Stacking combines many separate exposures of the same scene into a single image, and the reason it works comes down to a simple statistical fact: random noise partially cancels itself out when averaged across many frames, while the real signal — the actual stars, nebulae or landscape detail present in every frame — reinforces consistently, since it's genuinely there in each individual exposure rather than random.

The Math Behind Why It Actually Works

Random sensor noise in each individual frame is, by definition, uncorrelated from frame to frame — it doesn't land in the same place or at the same intensity twice. Averaging N frames together improves signal-to-noise ratio by a factor of the square root of N, a standard result from basic statistics rather than anything camera-specific. Stacking 4 frames improves signal-to-noise by a factor of 2 (roughly 2 stops); 10 frames by roughly 3.2x (about 3.3 stops); 100 frames by a full 10x (roughly 6.6 stops) — a genuinely enormous improvement, though with steeply diminishing returns per additional frame as the count grows, since doubling from 50 to 100 frames only adds about 1 stop, not another full doubling of benefit.

The Four Frame Types a Full Stacking Workflow Uses

Light frames are the actual exposures of the target — the stars, nebula or landscape being photographed. Dark frames are exposures of identical length and settings but with the lens cap on, capturing only the sensor's own thermal noise pattern so it can be subtracted out of the light frames. Bias frames are the shortest possible exposures with the lens cap on, capturing the sensor's baseline electronic readout noise independent of exposure length. Flat frames photograph an evenly lit blank surface (a twilight sky or a light panel) at the same aperture and focus as the lights, capturing lens vignetting and sensor dust spots so those can be corrected out too. A full calibrated stack uses all four frame types together; a simplified light-frames-only stack, common for wide-field Milky Way and star-trail work, skips darks, bias and flats and still gains the core signal-to-noise benefit from averaging alone.

Star-Field Stacking vs. Deep-Sky Stacking

Wide-field Milky Way stacking, the most common use for a general photographer, typically combines 10-30 exposures shot at each frame's own NPF-rule trailing limit, aligned and averaged in software to reduce noise while keeping stars sharp. Deep-sky stacking — targeting a specific nebula or galaxy through a telephoto lens or small telescope, usually with a tracker running — commonly combines dozens to hundreds of frames over multiple nights, since the target's actual signal is far fainter relative to noise than a broad star field and needs a much larger stack to pull out cleanly.

Star Alignment: Why Frames Need Registration Before Averaging

Because Earth rotates between exposures, even a well-timed sequence shows the stars in a very slightly different position from frame to frame unless a tracker is running. Stacking software registers — automatically detects matching stars and shifts each frame to align them precisely — before averaging, so the stacked result shows sharp, correctly aligned stars rather than a slightly smeared composite from unregistered frames. Any foreground element in an untracked sequence stays fixed relative to the frame edge while the stars shift underneath it, which is exactly why a separately exposed, unstacked foreground frame is usually blended in afterward rather than included in the star-aligned stack itself.

What Stacking Doesn't Fix

Stacking reduces random noise, but it does nothing for a genuinely missed focus point, a badly light-polluted location, or a poorly exposed individual frame — averaging together dozens of soft, out-of-focus frames just produces a soft, out-of-focus stack with less visible grain, not a sharp one. Every fundamental of a good single exposure covered elsewhere on this reference still has to be right before stacking adds its benefit; stacking amplifies a good starting point, it doesn't manufacture one from a bad frame.

Software Handles the Actual Averaging

Several dedicated stacking applications, along with general astrophotography processing software, handle registration, calibration-frame subtraction and averaging automatically once the frames are imported — the underlying math described above happens inside the software rather than requiring manual calculation for each stack, though understanding what it's actually doing makes it much easier to diagnose a stack that doesn't come out as expected.