Every before/after and scalp frame we have made

You were right that these were missing

174 frames

The complete archive listed them inside each revision's sheets, but they were never gathered, so there was no way to see the before/afters as a body of work. They are spread across 34 different revisions — v25 and v27 hold 16 each, growth-v18 and v38 hold 14 each, and v9 holds 12.

A filename search alone would have missed about fifty of them. Searching for ba, before, W07, W08 and the like found the obvious family. A visual sweep of every unscreened frame — 2,756 files, 747 new uniques — turned up whole families with no matching token in the name: the W45/W51/W61 length comparisons in v4/v5/v6, W71/W72 in v7, the R-series hair-diary Week 1 / Week 11 cards in v9 and diary-anonymous, growth-v18's B-series BEFORE|AFTER panels and its bare density and length plates, and the v23 scalp-parting frames.

449 duplicate source paths were collapsed — the same frame mirrored across hosted folders, re-hosted revisions and site copies. Each record on disk lists every path it was found at.

Two honest notes

v10 and v15/v16 contain none of these. I told you earlier those revisions were invisible to the first archive pass, which was true — but the search found zero before/after or scalp frames in v10's breakage/brush/identity folders or in v15/v16's angle-1…angle-5. They are 330 frames of something else.

There are genuine competitor before/after scalp images in the tree — Georgemane, Divi, Act+Acre — sitting in the v9/v10 hosted research folders and obvi-v17's references. Those are scraped competitor assets, not ours, so they are deliberately not in this set. Say the word and I'll stage them as a separate reference group; they are the only real photographic scalp results in the whole tree.

Full accounting of near-misses and deliberate exclusions: accounting.md

By revision — most frames first