Why does stacking not just add SNR linearly?

When you average N sub-frames, the random noise falls by the square root of N while the signal stays constant. So the stacked SNR is the single-sub SNR multiplied by the square root of the number of subs, not by N itself.

How many subs do I need to double my SNR?

Because SNR grows with the square root of frame count, doubling your SNR requires four times as many subs. Going from 10 to 40 frames doubles SNR; going from 40 to 160 doubles it again. This is the diminishing-returns reality of long integrations.

What is a good target SNR?

It depends on the target and how much you stretch the image. Bright objects look clean at an SNR around 20, while faint outer nebulosity that you stretch hard can need an SNR of 50 or more before noise stops being visible.

Does read noise matter here?

Read noise sets the SNR of a single sub for a given exposure length. Longer subs collect more signal relative to a fixed read-noise floor, raising single-sub SNR. This tool takes your measured single-sub SNR as the input, so the read-noise effect is already baked in.

Can I trade longer subs for fewer frames?

Yes. A longer sub raises the SNR of each frame, so you need fewer of them. But very long subs risk losing frames to satellites, wind, or guiding errors. Many imagers pick a sub length that comfortably swamps read noise, then stack enough of them.

Astrophotography Total Exposure Calculator

In deep-sky astrophotography, the final image quality is driven by total integration time, but the relationship between the number of sub-frames you stack and the resulting smoothness is not linear. This calculator uses the quadrature stacking rule to tell you how many subs and how much total time you need to reach a target signal-to-noise ratio.

How it works

Stacking N sub-frames averages out random noise, which falls as the square root of N while the signal is unchanged. The stacked SNR is therefore:

SNR_stacked = SNR_single × √N

Solving for the number of subs needed to reach a target SNR gives:

N        = (SNR_target / SNR_single)²
total_t  = N × sub_length

Because of the square, reaching a higher target costs disproportionately more frames — doubling SNR needs four times the subs and four times the total time.

Example and tips

If a single 3-minute sub yields an SNR of 8 and you want a final SNR of 40, you need (40 / 8)² = 25 subs, or 75 minutes of integration. To push that to SNR 80 you would need 100 subs and five hours. This is why faint targets demand whole nights of data. Keep individual subs long enough to swamp read noise, but short enough that you can discard the occasional ruined frame without losing much.