Why this is nearly free
Once frames are registered they share a coordinate system, so a fixed aperture at a fixed position measures the same star in every one of them. No re-integration is needed and the core pipeline is not touched — these tools only read what integration already wrote. That is also their limit: they report candidates and sessions, not calibrated photometry.
Three things the frames still know
Each one is honest about what fools it — that part matters more than the feature.
Session replay
Watch the stack accumulate: the cumulative image, the measured signal-to-noise curve as frames are added, and a timeline of which subs were accepted and which were thrown out.
Variability search
A light curve per star, judged as excess scatter over the session's own noise-versus-brightness trend. Nebula background is the main liar here — on one real field it produced eight of nine false candidates — so the test is stated relative to the field's own stars.
Gaia cross-check
Flags sources with no Gaia counterpart. On public frames of M101 it recovered SN 2023ixf to within 0.4 arcseconds. Cosmic ray tracks are narrower than the point spread function, so they are caught by shape rather than size — and "absent from Gaia" still does not mean "new".
A note on false positives
Every one of these produced convincing fakes first
A variability search is mostly a fight against things that look like discoveries. Noise peaks near zero flux, saturated cores that lose light as seeing changes, and above all nebulosity in the sky annulus — which moves the measured background around with the seeing.
Each of those was found by looking, not by theory, and each is gated now. The gates are written relative to the field itself, so a clean field excludes nothing and there is no per-target tuning to get wrong.
Request beta access