Project analysis
Frame measurements & integration diagnostics
Compare frames within one pointing and filter.
Processing outcomes describe participation, not image quality.
Integration depth
Measured noise against an independent-frame comparison
Frame health
Inspect variation; use combined metrics to assess quality
Frame measurements
Open a filename for its path, registration details and advisories.
| Frame / details | Measurement | Depth | Exposure s | Stars | HFR px | FWHM px | Eccentricity | Peak SNR | Noise |
|---|
Processing time
What the timings include
Wall-clock seconds, including failed attempts. Measurement includes conversion, quality detection, PSF fits and native noise/background. Registration includes its own star detection. Sampling includes normalization and temporary writes. Total includes overhead, but excludes report rendering and writing. Stage times do not overlap.
Method & interpretation
Advisories never automatically exclude a frame. A high SNR or star count alone is not a quality verdict.
Assumptions & limitations
Full-depth analysis uses non-overlapping 16×16 patches at two-pixel pitch in the central 80%. Only patches valid in every depth-included frame contribute. Noise is median patch MAD × 1.4826 of (odd sum − even sum)/N; at least 32 common patches are required. Odd depths have no split measurement. Shared systematic errors cancel; robust noise describes the distribution core, not total outlier energy.
The ideal 1/√N comparison assumes independent, equal-quality frames and is anchored at two frames. Trend fits use geometric even-depth checkpoints over the last fourfold window. Cumulative observations are correlated: R² is descriptive, not a confidence interval. Any conditional estimate assumes the fitted trend persists.