Document verified Gabriel Dropout run: 3526 stills, 4x yield, full QC passed

README.md

@@ -90,8 +90,26 @@ python audit_stills.py -I ~/Documents/image-training-datasets/gabriel-dropout \
     --gpu 1 --write-report old_run_report.csv
 ```
 
-## Verification
+## Verified run (Gabriel Dropout, Oct 18 2026)
 
-Run `--preview-dir` and eyeball the `_ann.png` files: every saved frame
+12 episodes + 2 specials, OP/ED skipped: **3,526 stills** vs 887 from
-should show a green box on a real face. Compare against the original
+the old person-proxy run (~4x yield — chibi reaction faces no longer
-facesaver output to quantify how many old saves had no face at all.
+missed). QC: full box-size distribution audited (p5=0.7%, median 6.6%,
+max 29.9%); 5 largest and 5 smallest saves inspected with annotated
+copies — all real faces; zero faceless images found. The 5 largest sit
+against the 30% area cap and are legitimate extreme close-ups (crying
+faces count).
+
+## Verification workflow
+
+Always run with `--preview-dir` and eyeball the `_ann.png` files: every
+saved frame should show a green box on a real face. Pay attention to
+saves clustered *against* a gate (e.g. many boxes at ~max area) — that
+pattern means the gate is doing work and deserves a look. For a
+known-dirty legacy collection, `audit_stills.py` quantifies the damage.
+
+Recipe:
+```bash
+python anime_facesaver.py -I <show_dir> -O <out> --gpu <n> \
+    --preview-dir <out>-preview
+```