The survey footprint is being built as a positive coverage map minus the healsparse mask bits (#845, #847). One masking input never leaves the pixel domain: the instrument flag images (bad columns, saturated pixels, bleed trails). The coverage map of #797 is built from CCD corner WCS in the headers, so it cannot subtract them — the footprint would silently include defective pixels.
The lost area is percent-level, but it carries exactly the complex small-scale geometry (thin bad columns, bleed trails) that an accurate window function needs.
Proposal: emit the instrument flags as a healsparse defect map — per exposure, footprint ∧ flags rasterized at the mask nside, then combined across epochs the same way #797 combines coverage. This is a natural extension of #797's map builder, which already walks the exposure headers; only ShapePipe reads these flag files, so the map has to come from our side.
— Claude on behalf of Cail
The survey footprint is being built as a positive coverage map minus the healsparse mask bits (#845, #847). One masking input never leaves the pixel domain: the instrument flag images (bad columns, saturated pixels, bleed trails). The coverage map of #797 is built from CCD corner WCS in the headers, so it cannot subtract them — the footprint would silently include defective pixels.
The lost area is percent-level, but it carries exactly the complex small-scale geometry (thin bad columns, bleed trails) that an accurate window function needs.
Proposal: emit the instrument flags as a healsparse defect map — per exposure, footprint ∧ flags rasterized at the mask nside, then combined across epochs the same way #797 combines coverage. This is a natural extension of #797's map builder, which already walks the exposure headers; only ShapePipe reads these flag files, so the map has to come from our side.
— Claude on behalf of Cail