How PiFinder measures sky quality
PiFinder's plate-solving camera doubles as a sky-quality meter. The three techniques behind it — and how well it tracks a hand-held SQM-L, with an interactive chart of every test session.
PiFinder finds objects in the night sky by photographing the stars and recognising the pattern (plate solving). The same camera doubles as a sky-darkness meter: it continuously reports how dark your sky is on the SQM scale. Three techniques work together — a pixel-brightness measurement produces the number, a colour trick corrects it on very dark skies, and star brightnesses act as an independent watchdog. Readings from a hand-held SQM-L meter, taken side by side, are the ground truth everything below is tested against. Dotted words show an explanation when you hover.
Techniques
Background brightness the published number
Most pixels in a sky photo contain no stars — just sky. We take the middle brightness of those empty pixels, subtract the camera's own electronic glow, and divide by how long the shutter was open. A factory calibration converts the result to the standard scale. Readings from the last 15 seconds are combined and published once per second. Because no stars are needed, it keeps working under cloud and while the telescope moves.
Airglow correction dark-sky fix
Even a perfect night sky glows faintly: the atmosphere emits its own light (airglow), mostly in the infrared. The imx462 has no IR filter, so it sees far more of that glow than the hand-held meter does — its dark-sky readings come out too bright. Airglow has a tell, though: it registers equally in the camera's red, green and blue colour filters, while ordinary skylight lands mostly in the green pixels. The extra red in the background therefore measures the airglow, and we subtract it. The HQ camera needs far less of this: its built-in IR filter blocks the airglow the same way the SQM-L's filter does.
Star check watchdog — not an input
When plate solving succeeds, we compare how bright each star looks against its star catalogue value. This never feeds the published number directly. It answers a different question: is something dimming the view? Stars fading while the sky background stays put means cloud — the reading is flagged unreliable and self-calibration pauses. If the fading pattern instead looks like dew or dirty optics under a clear sky, the measured light-loss is subtracted from the published value until the optics recover.
Cameras
| Camera | Type | Sees infrared? | Glow reference | Airglow fix | Star check |
|---|---|---|---|---|---|
| imx462 | colour | yes — no IR filter | shielded pixels, every frame | needed — calibrated | yes |
| HQ / imx477 | colour | no — IR filter built in | factory value + self-learning | barely needed | yes |
| imx296 | mono | partially | factory value + self-learning | impossible — needs colour | yes |
The infrared column is the key difference. The HQ camera's IR filter gives it nearly the same spectral window as the SQM-L reference meter, so its dark-sky readings stay close without correction; the unfiltered imx462 sees the infrared airglow in full and relies on the colour-based fix above. A mono camera can't see the colour signature airglow leaves, so it runs the plain background measurement — the imx296 also has only a single moonlit reference session behind it and is the least-tested profile.
Performance — replay of 53 test sessions
Each dot is one test session: PiFinder's reading minus the hand-held meter's, placed left-to-right by how dark the sky was. On brighter skies the two versions agree, both inside the ±0.10 band. Past 20 mag the uncorrected reading drifts up to a full magnitude too bright, while the airglow fix holds the error to roughly ±0.1–0.2 mag — which is why the fix is what your PiFinder actually runs. Hover any dot for its session. Data: July 2026 replay of PiFinder's full calibration archive.