diff --git a/docs/re/BACKLOG.md b/docs/re/BACKLOG.md
index 6c1932e..f32db75 100644
--- a/docs/re/BACKLOG.md
+++ b/docs/re/BACKLOG.md
@@ -6499,3 +6499,34 @@ sets, abandoned for the shared `test` table" is a reading, not a measurement.
 116 of 130 referenced values are bare `Equip_EnumGameModel_wep_NN.tbl` names from
 `PlayerLOD`/`WingmanLOD`/`PlayerMotion`/`WingmanMotion` (58 distinct each); only
 the 14 stage-tagged values use the `+` form.
+
+---
+
+## ✅❌ 2026-08-28 — challenge LOD tables are `_test` + set dressing; orphan count corrected to 6
+
+Item (a).
+
+✅ **The five bespoke challenge tables are exact supersets of `_test`** — 10 of 10
+(5 stages × 2 families) with **zero removals**; extras are +2 to +9 and name the
+stage's own scenery (`stg24_01…04`, `stg26_01…03`, `stg27_01…02`, `stg29_01…02`)
+plus `rot_n001_break`.
+
+🔑 **S28's nine extras independently confirm the player-craft result**:
+`LOD_Frame_rou_f004`, four `rou_f004_mnv*`/`turn180` poses and three `eff_j004_*`
+effects — the LOD table for the one stage that flies `rou_f004`.
+
+⚠️ This does **not** revive "`rot_n001` is on the disc": that refutation concerned
+the **bare** name (still 0 exact); only the `_break` form occurs, as recorded.
+
+❌ **Self-correction — 6 orphan tables, not 8.** `name_hash` is case-**insensitive**,
+so `_test` and `_Test` are the *same pak entry*
+(`EnumLODSet` → 3485720498, `EnumGameModel` → 4020329537 for both spellings). The
+family has **10 distinct keys**, and the unreferenced set is **S01, S02, S16 × 2
+= 6**.
+
+🟡 **The six are not stale copies.** They are far smaller (120/129/178 vs 676
+`EnumLODSet` fields) yet each holds **4–5 entries `_test` lacks** —
+`LOD_Effect_eff_e0033/_e0058/_e0059/_e0060/_n0071`, and S16 adds five
+`rou_e106_*` boss parts (cf. `stage16-boss`). The live table is **not** a
+superset of the abandoned ones. `Tutorial` has the same shape: 66 fields, 3
+unique.
diff --git a/docs/re/structures/stage-mission-tables.md b/docs/re/structures/stage-mission-tables.md
index 9658251..5d7168e 100644
--- a/docs/re/structures/stage-mission-tables.md
+++ b/docs/re/structures/stage-mission-tables.md
@@ -322,3 +322,49 @@ per-stage LOD sets, abandoned when the story campaign was pointed at the shared
 bare `Equip_EnumGameModel_wep_NN.tbl`-style names reached from `PlayerLOD`,
 `WingmanLOD`, `PlayerMotion` and `WingmanMotion` (58 distinct each). Only the 14
 stage-tagged values use the `+` path form.
+
+### ✅ Each challenge LOD table is `_test` **plus its own set dressing**
+
+Diffing the field sets — the same move that cracked `EnumWeapon_EX5` — the five
+bespoke challenge tables are **exact supersets of `_test`, 10 of 10 with zero
+removals** (five stages × the `EnumLODSet` and `EnumGameModel` families):
+
+| stage | `EnumLODSet` extras | `EnumGameModel` extras |
+|---|---:|---:|
+| S24 | +5 | +5 |
+| S26 | +4 | +4 |
+| S27 | +3 | +3 |
+| S28 | **+9** | +5 |
+| S29 | +2 | +2 |
+
+🔑 **The extras name the stage's own scenery** — `…_stg24_01…04`, `…_stg26_01…03`,
+`…_stg27_01…02`, `…_stg29_01…02` — plus `rot_n001_break`, shared by S24, S26, S27,
+S28 and Tutorial.
+
+🔑 **S28's extras independently confirm the player-craft finding.** Its nine are
+`LOD_Frame_rou_f004`, four `rou_f004_mnv/turn180` motion poses, three
+`eff_j004_f0N` effects and `rot_n001_break` — i.e. **the LOD table for the one
+stage that flies `rou_f004`** carries exactly that craft and its poses
+([[stage-numbering-and-player-craft]]). Two unrelated tables agreeing on S28's
+oddity is a real cross-check.
+
+⚠️ `rot_n001_break` here does **not** revive "`rot_n001` is on the disc" — that
+refutation was about the **bare** name, which still occurs 0 times exactly; only
+the `_break` form is real, exactly as recorded.
+
+### ❌ Correction: **6** orphan tables, not 8 — `name_hash` is case-insensitive
+
+The previous pass counted `_Test` and `_test` as two tags. They are **one pak
+entry**: `name_hash` is case-insensitive, so
+`name_hash("EnumLODSet_test.tbl") == name_hash("EnumLODSet_Test.tbl")` —
+**3485720498** for both, and **4020329537** for the `EnumGameModel` pair. The
+family therefore has **10 distinct keys**, not 11 tags, and the unreferenced set
+is **S01, S02, S16 × 2 = 6 tables**, not 8.
+
+🟡 **And the six are not stale copies of `_test`.** They are far smaller — 120,
+129 and 178 `EnumLODSet` fields against `_test`'s 676 — yet each carries **4–5
+entries `_test` does not have**: `LOD_Effect_eff_e0033`, `_e0058`, `_e0059`,
+`_e0060`, `_n0071` (and S16 adds five `rou_e106_*` boss parts, matching
+[[stage16-boss]]). So the live table is not a superset of the abandoned ones;
+whatever those effect entries were for, the shipped campaign table drops them.
+`Tutorial` is the same shape — 66 fields, 3 of them unique.