KOPR DOCUMENTATION BASELINE
Revision 10.22 — 27 July 2026
Current documented application line: R81 DSLR RAW provenance and WCCD viewport hotfix

R81 STATUS
- R81 is a focused hotfix release over R80/R79 and changes no scientific algorithm.
- Complete deterministic QA: 1231 collected; 1225 passed; 6 skipped; 0 failed.
- R80 focused integration: 24 passed. R81 focused gates: 31 RAW/materialization and 45 redraw/integration tests passed.

R75 CALIBRATION SAFETY
- Master-flat correction is guarded/confirmed before build commitment.
- Stack Parameters performs one preflight pass before execution.
- These guards do not change approved calibration or stacking mathematics.

R76 CAMERA, BINNING, SCALE AND LIGHT-TIME
- Camera identity/native geometry is explicit and separate from validated observation binning.
- Shared physical-scale priority: solved WCS or trusted online calibration; manual override; FITS scale metadata; active optics with validated binning.
- Internal DE440s comet positions apply down-leg light-time.
- Guide/Tycho/Horizons comparisons are diagnostic; no universal arcsecond acceptance limit is imposed.

R77–R78 REFERENCE STARS
- AutoStars and Manual Stars use one lossless canonical type/collection.
- Extinction is applied to q_i before selection.
- A global solver selects a strict mutually consistent subset; every pair must differ by < 0.150 mag, so exactly 0.150 mag fails.
- user_enabled is separate from pairwise_accepted.
- Reference stars, Process obs. and Af-rho use one accepted-ID set and one final solution.
- No parallel post-success scatter gate; internal pairwise audit text is not added to ICQ comments.

RETAINED CONTRACTS
- R74 automatic DSLR RAW → TG FITS materialization and FITS-only session.
- R73 shared saturation/camera workflow.
- R61 cumulative centroid/settings/comment invariants.
- R55 Automatic Star Measure crowded-field stability.
- R48 blind-online/fixed-offline astrometry; R47/R40 saturation; R45 robust solve; R29 GUI; R27 photometry; R22 performance/multiplatform.
- Hidden complete Automatic measure workflow remains unpublished. Official releases and Czech WSL PDF remain removed.

OUTSTANDING TARGET TESTS
- Real master-flat build with calibration frames and interactive Stack Parameters responsiveness.
- Rendered Reference stars dialog with many stars.
- Real DSLR decoding and camera-dialog suppression.
- Real-image down-leg light-time residual comparison.
- End-to-end Process obs. and Af-rho.
- Archival five-observation photometry field validation remains NOT RUN.


R80 REFERENCE STARS USE STATE
- Use represents effective inclusion, not raw user_enabled state.
- Effective check: user_enabled AND quality_eligible AND pairwise_accepted is not False.
- Pairwise/quality/user exclusions display unchecked; Reason is the sole visible explanation.
- Pairwise status column is absent. Internal observer/solver states remain lossless and separate.
- A rechecked candidate stays selected while pending; Recalculate reapplies strict policy.

R81 RAW PROVENANCE AND VIEWPORT
- Direct TG FITS: RAWREP=T. Calibrated/astrometric derivatives: RAWREP=F.
- Final stacks remove RAWREP, RAWFILE, RAWEXT, RAWMODE, RAWVERS, RAWSIZE and RAWMTIME; aggregate ancestry remains in HISTORY, REFIMAGE, NCOMBINE and PROVENANCE.
- Discovery ignores historical RAW claims on derived products, supports legacy valid _TG.fits without RAWREP, and still blocks two genuine direct representations.
- Repeated comet/star/tail measurement preserves viewport when the required image role is already active; role changes still redraw once.