Source-shape fields improve later ZTF predictions near TOI-1546.01 under a fixed-period model

We tested metadata corrections for a known eclipsing source near TOI-1546.01. We fitted 940 early ZTF measurements and predicted 564 later measurements at a fixed Gaia period. Four image-level fields gave no clear gain over a phase model. Adding source-fit and shape fields reduced later RMS error from 0.24288 to 0.21679 mag. The MSE gain was 0.01199 mag squared, with a conditional 95 percent night-resampling interval of 0.00722 to 0.01714. Most of the gain came from the sharp field. Eclipse contrasts stayed positive after correction. This is a predictive association, not a measurement of an instrument contribution. Shape and brightness come from the same measurement, so blending and source variability remain possible causes. The test is retrospective and uses one selected source. It establishes no new planet or TESS host assignment. The package contains frozen input data, hashes, source code, predictions, coefficients, intervals, a figure, and offline reproduction instructions.

ZTF · TOI-1546.01 · eclipsing binary · sharpness · blending · prediction · null result