Paper 1 finish plan

Two active research workstreams → one acceptance pilot → one final recipe → five-field validation → manuscript.

Finish by demonstrating useful improvement within a stated domain. Do not require zero residuals or a unique physical explanation of every feature.

Work package	Owner	Status	Finish condition
1 · Neighbour ledger	Trace PS1 star removal records	ACTIVE	Exact source/mask/seam associations; full versus partial removal; say whether pixels changed.
2 · Neighbour-aware fit	Summarize ePSF atlas findings	PILOT COMPLETE; FOLLOW-UP READY	Fixed hp_d first; matched baseline, source roles and held-out evidence. Reuse existing joint-fit work.
3 · Background and recipe	User + coordination	OPEN	Judge on the neighbour-aware scene; make only the targeted comparison still needed.
4 · Final validation	Coordination	PENDING PILOT	Convergence, alternate starts, K=5, astrometry/flux/profile errors and residual grids.
5 · Per-band and extraction	Validate faint-star photometry + coordination	ACTIVE / FINAL INPUTS PENDING	Seam bounds, flux-sum checks, matched Hotpants control, refit stability, exact faint-star add-back.
6 · Figures and manuscript	Existing publication workstream + coordination	STRUCTURE NOW	Build the draft around claims; replace provisional figures only after final data validation.

Important distinctions

Ledger or scene changes do not automatically require rebuilding template pixels.
Nuisance sources model contamination; ePSF contributors train the shared profile. Record the roles explicitly.
Partially retained sources must account for light already subtracted through the template.
Improved neighbour modelling does not itself validate targeted faint-star photometry.
Keep A3, lp7s and no width term as the baseline unless a specific failed validation requires a change.

Finish the existing ledger and neighbour-fit pilots; prepare paper structure in parallel.
Accept a versioned handoff; run the F1 pilot and already-planned independent comparison.
Settle the background and freeze one recipe.
Run final five-field products and independent validation; complete faint-star extraction.
Replace provisional figures, complete the manuscript, check citations and reproducibility, then coauthor review and submission.

Paper 2 temporal performance does not block Paper 1. No new broad campaign or task reassignment was launched by this plan.