# D3D-003-LAB1 — learner guide

Goal: build `Capture-to-Reconstruction Method Comparison and Evidence Pack v1` from visible synthetic evidence.

1. Copy `evidence/evidence_pack.template.json` to a new file.
2. Populate it from `fixtures/baseline_observations.json`, including `camera_count`, `sparse_point_count`, provenance status, and traceable per-method evidence references.
3. Write your own `decision.selected_path`, `decision.final_state`, `decision.target_constraint_scope`, and rationale. `SELECTED` is for the standard declared target; `TARGET_SPECIFIC` requires `CONSTRAINED_NAMED_TARGET_ONLY`.
4. Run `python run_checks.py evidence/<your-file>.json`.
5. Repeat with `mutation_low_overlap.json`, `mutation_metric_editable_mesh.json`, and `mutation_constrained_hardware.json`. Do not copy the baseline decision.
6. For S4, keep `transfer/transfer_fixture.json` as the observation source and copy `transfer/transfer_evidence_pack.template.json` to a new submission file. Fill only the decision, rationale, and the five `decision.triggers` (`review`, `migration`, `escalation`, `stop`, `pivot`) from the changed-context evidence; do not copy the baseline decision.
7. Run `python run_checks.py transfer/<your-transfer-pack>.json` before submitting S4.

The checker validates canonical consistency only; it does not choose a method for you. Keep missing evidence visible instead of inventing it. Use only synthetic/self-created/public-domain/rights-cleared captures.