KOPR SCIENTIFIC / COMPACT FITS QUICK REFERENCE
Baseline: KOPR_SCIENTIFIC_COMPACT_VALIDATION_BATCH10_BUILD_20260724

PURPOSE
Scientific and Compact change only the final on-disk representation.
All calibration, reprojection, stacking, manual WCCD and automatic-helper calculations remain floating point.

SETTINGS
Settings -> Processing -> FITS output format
0 = Scientific; 1 = Compact (settings.dat second positional row, index 14)
Old configurations default to Scientific. Use GetFitsOutputMode()/GetFitsOutputModeIndex().

SCIENTIFIC
BITPIX=-32, KOPRMODE=SCIENTIFIC, STORFMT=FLOAT32, QSTEP=0
Invalid pixels remain NaN.

COMPACT
BITPIX=16, KOPRMODE=COMPACT, STORFMT=INT16, BLANK=-32768
physical = stored * BSCALE + BZERO
QSTEP = BSCALE = (DATAMAX-DATAMIN)/65534 for non-constant data
Finite codes: -32767..32767

SAFETY
Accept only when QSTEP <= 0.25 * sigma_local.
Noise estimate: robust median/MAD of horizontal and vertical neighbour differences.
Unsafe products fall back independently to Scientific; no clipping or coarse hidden quantization.
Policy cards: REQMODE QPOLICY QMETHOD QFRAC QNOISE QMAX QNSAMP QSAFE QFALLBK
Actual storage cards: KOPRMODE STORFMT BITPIX QSTEP

AFFECTED PRODUCTS
Created/imported calibration masters, calibrated LIGHTs, Stars stacks, Comet stacks and joint pairs; MEAN, SUM and SIGCLIP methods.

STACK TIME
DATE-BEG first accepted start
DATE-OBS/DATE-MID/DATE-AVG exposure-weighted midpoint
MJD-OBS/MJD-AVG same midpoint in MJD UTC
DATE-END last accepted end
TELAPSE includes gaps; TIMEALGO=EXP-WMID; TIMECOMP=COMPLETE/PARTIAL

EXPOSURE
NCOMBINE actual frames; EXP_NR compatibility alias
TOTEXP=XPOSURE=sum accepted exposure times
MEAN/SIGCLIP EXPTIME=one input exposure; SUM EXPTIME=total
WCCD uses net_flux/EXPTIME after additive local-sky subtraction
Matched Comet/Stars EXPTIME tolerance=0.001 s; accepted value freezes at preflight
TOTEXP/XPOSURE remains reporting metadata

INTEROPERABILITY
Astrometry preserves primary storage and all extensions; no second Compact quantization.
WCCD decodes supported integer/float FITS to physical float32.
Native Scientific and Compact products use identical decoded physical arrays for WCCD and helper calculations.
External FITS can be measured manually; automatic hand-off still requires the full native WStack pair/QC/provenance contract.

VALIDATE
python validate_scientific_compact.py --json validation.json --csv validation.csv --markdown validation.md
python validate_scientific_compact.py --compare scientific.fits compact.fits --json pair-validation.json
python validate_scientific_compact.py --source input.fits --output-dir validation-products --keep-files