BT410 5.3.89: the hit-location cylinder MEASURED -- 18 tables, 8 distinct, and two chassis that never twist

The operator recalled the damage model as "a pie wedged cylinder" and asked
how it maps across the mechs.  It is exactly that, and the shipped data is now
extracted rather than described.

dmgscan.py brute-forces every offset in BTL4.RES and accepts a candidate only
if the ENTIRE nested type-29 structure parses -- thresholds strictly ascending
and terminating at exactly 1.0, zone indices in range, names NUL-terminated.
A wrong format guess cannot survive that, so finding exactly 18 tables -- the
count DAMAGE-MODEL.md already claimed from an independent reversal -- is a
confirmation of the format, not a coincidence.

MEASURED: 18 tables, every one 7 bands x 8 wedges = 56 cells.  Only EIGHT are
distinct by content; the other ten are duplicates.

  22 zones  4 twisting  x3  Avatar / Mad Cat class -- table A
  22 zones  4 twisting  x2  Avatar / Mad Cat class -- table B
  21 zones  4 twisting  x3  Loki
  22 zones  4 twisting  x2  Thor
  21 zones  4 twisting  x2  SND2
  24 zones  4 twisting  x2  Battlemaster / Vulture
  20 zones  0 twisting  x2  Black Hawk
  17 zones  0 twisting  x2  Owens

BLACK HAWK AND OWENS ROTATE NO BAND WITH THE TORSO.  Every other chassis
rotates its upper four.  That is a real behavioural difference in the shipped
data, not an absence of it.

The zone COUNTS match the per-chassis .SKL dz_ sets exactly, which is what
lets a table be fingerprinted back to a chassis.  It is not always unique --
Avatar and Mad Cat share a zone set but have two DIFFERENT tables, and no
chassis name sits near the stream, so they are recorded A/B rather than
guessed.  Stated as undetermined in both the notes and the visual.

THE GEOMETRY, now named: 18 wedge names in six anatomical rings (Foot, Leg,
Hip, Waist, Chest, Top).  Slot 0 starts at angle 0 spanning 45 degrees, so
under atan2(z,x) the mech's +X is right and +Z is front.  Each named face
covers TWO adjacent wedges (Right = 7,0 / Front = 1,2 / Left = 3,4 / Rear =
5,6) -- so dead ahead is the SEAM between two Front cells, never the centre of
one.  Bands 0-2 are chassis-fixed; the live torso twist is added to the impact
angle before the wedge pick on the rest.

And the scatter is generous in a way worth knowing at the controls: a clean
foot-wedge hit is only 50% that foot, 30% the lower leg, and 20% of the time
the OTHER foot entirely.

ALSO: an interactive plate of all of it -- every cell of all 8 tables, plan and
elevation, and a twist slider that rotates the upper bands live -- published
for the playtesters.  Its dataset regenerates from dmgscan.py.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

restoration/dmgscan.py

@@ -0,0 +1,105 @@
+#
+# dmgscan -- extract the type-29 cylinder hit-location tables from BTL4.RES.
+#
+# The unaimed-damage model is a cylinder around the mech: 7 height BANDS x 8
+# angular WEDGES, and each of the 56 cells carries its own cumulative
+# distribution over that chassis's armor zones.  DMGTABLE.CPP resolves a hit
+# by height -> band, atan2(z,x) -> wedge, then one uniform roll down the
+# cell's table.
+#
+# The stream format is documented in source410/BT/DMGTABLE.CPP and this
+# scanner is the check on it: it brute-forces every offset in the resource and
+# only accepts a candidate whose ENTIRE nested structure parses -- thresholds
+# strictly ascending and terminating at exactly 1.0, zone indices in range,
+# names NUL-terminated.  A wrong format guess cannot survive that, which is
+# why finding exactly 18 tables (the count the notes already claimed, from an
+# independent reversal) is a real confirmation rather than a coincidence.
+#
+# Findings, 2026-07-30:
+#   * 18 tables, every one 7 rows x 8 cells.
+#   * Only 8 are DISTINCT by content; the rest are duplicates.
+#   * Bands 0-2 are chassis-fixed, bands 3-6 rotateWithTorso -- except on
+#     Black Hawk and Owens, where NO band rotates.
+#   * 18 wedge names in 6 anatomical rings: Foot, Leg, Hip, Waist, Chest, Top.
+#   * Zone counts (17/20/21/22/24) match the per-chassis .SKL dz_ sets exactly,
+#     which is what lets a table be fingerprinted back to a chassis family.
+#
+# Usage:  python3 dmgscan.py  [path/to/BTL4.RES]
+#
+# Format (DMGTABLE.CPP, byte-verified against the real streams):
+#   Table { i32 rowCount;  Row[rowCount] }
+#   Row   { i32 rotateWithTorso;  i32 cellCount;  Cell[cellCount] }
+#   Cell  { i32 nameLen; char name[nameLen]; u8 0; i32 entryCount; Entry[] }
+#   Entry { f32 cumulativeThreshold; i32 zoneIndex }
+#
+import struct, sys, json
+
+d = open(sys.argv[1] if len(sys.argv) > 1 else r'../ALPHA_1/REL410/BT/BTL4.RES', 'rb').read()
+N = len(d)
+u32 = lambda o: struct.unpack_from('<I', d, o)[0]
+i32 = lambda o: struct.unpack_from('<i', d, o)[0]
+f32 = lambda o: struct.unpack_from('<f', d, o)[0]
+
+def cell(o):
+    if o + 4 > N: return None
+    ln = i32(o)
+    if not (0 <= ln <= 64): return None
+    o2 = o + 4
+    if o2 + ln + 1 > N: return None
+    nm = d[o2:o2+ln]
+    if ln and not all(32 <= c < 127 for c in nm): return None
+    if d[o2+ln] != 0: return None
+    o2 += ln + 1
+    if o2 + 4 > N: return None
+    ec = i32(o2); o2 += 4
+    if not (1 <= ec <= 40): return None
+    if o2 + 8*ec > N: return None
+    ents, prev = [], -1.0
+    for k in range(ec):
+        t = f32(o2); z = i32(o2+4); o2 += 8
+        if not (0.0 < t <= 1.0001) or t < prev - 1e-6: return None
+        if not (0 <= z < 64): return None
+        ents.append((t, z)); prev = t
+    if abs(ents[-1][0] - 1.0) > 1e-3: return None
+    return nm.decode('ascii', 'ignore'), ents, o2
+
+def row(o):
+    if o + 8 > N: return None
+    rt = i32(o); cc = i32(o+4)
+    if rt not in (0, 1): return None
+    if not (1 <= cc <= 32): return None
+    o2 = o + 8; cells = []
+    for _ in range(cc):
+        c = cell(o2)
+        if c is None: return None
+        cells.append((c[0], c[1])); o2 = c[2]
+    return rt, cells, o2
+
+def table(o):
+    if o + 4 > N: return None
+    rc = i32(o)
+    if not (2 <= rc <= 32): return None
+    o2 = o + 4; rows = []
+    for _ in range(rc):
+        r = row(o2)
+        if r is None: return None
+        rows.append((r[0], r[1])); o2 = r[2]
+    return rows, o2
+
+out, o = [], 0
+while o < N - 8:
+    t = table(o)
+    if t and len(t[0]) >= 3 and sum(len(r[1]) for r in t[0]) >= 12:
+        out.append((o, t[0])); o = t[1]
+    else:
+        o += 1
+print("tables found: %d" % len(out))
+for off, rows in out:
+    cells = sum(len(r[1]) for r in rows)
+    rot = sum(1 for r in rows if r[0])
+    zs = sorted({z for r in rows for _, es in r[1] for _, z in es})
+    print("  @%#08x  %d rows x %d cells (%d)  torso-rotating rows: %d  zones used: %d"
+          % (off, len(rows), len(rows[0][1]), cells, rot, len(zs)))
+json.dump([{'off': o, 'rows': [{'rot': r[0], 'cells': [{'name': c[0], 'entries': c[1]}
+           for c in r[1]]} for r in rows]} for o, rows in out],
+          open('dmgtables.json', 'w'))

restoration/source410/BT/DAMAGE-MODEL.md

@@ -195,3 +195,67 @@ untouched PPC_1 keeps shooting.
 - Leg-branch gates use "not already destroyed" instead of the live
   MovementMode/IsDisabled checks (gait FSM pending).
 - Gyro hit-feed and destroyed-skin graphics are log stubs (feel/render waves).
+
+## 11. The cylinder table, measured — `dmgscan.py`
+
+Section 4.3 describes the unaimed resolve; this is the shipped data behind it,
+extracted by `restoration/dmgscan.py` (brute-forces every offset and accepts
+only candidates whose entire nested structure parses, so the format itself is
+under test). **18 tables, every one 7 bands x 8 wedges = 56 cells.**
+
+**Only 8 are distinct by content** — the other 10 are duplicates.
+
+| distinct table | zones | torso-rotating bands | copies | chassis family (by zone-set fingerprint) |
+|---|---|---|---|---|
+| `b586e6f9` | 22 | 4 | 3 | Avatar / Mad Cat class — table A |
+| `bdf3d3a8` | 22 | 4 | 2 | Avatar / Mad Cat class — table B |
+| `ad4eb428` | 21 | 4 | 3 | Loki |
+| `db6cc00d` | 22 | 4 | 2 | Thor |
+| `2b34fe3e` | 21 | 4 | 2 | SND2 |
+| `9b481293` | 24 | 4 | 2 | Battlemaster / Vulture |
+| `11c840bc` | 20 | **0** | 2 | Black Hawk |
+| `4c07e216` | 17 | **0** | 2 | Owens |
+
+The zone COUNTS (17/20/21/22/24) match the per-chassis `.SKL` `dz_` sets
+exactly, which is what allows the fingerprint. It is not always unique: the
+Avatar and Mad Cat class share a zone set but have two DIFFERENT tables, and
+the resource carries no chassis name near the stream, so which is which is
+undetermined — recorded as A/B rather than guessed.
+
+**Black Hawk and Owens rotate NO band with the torso.** Everything else
+rotates its upper four. That is a real behavioural difference, not missing
+data.
+
+### The 18 wedge names — six anatomical rings
+
+```
+band 6  TopRight   TopLeft                              (2 names / 8 slots)
+band 5  RightChest FrontChest LeftChest RearChest       (4)
+band 4  RightWaist FrontWaist LeftWaist RearWaist       (4)
+band 3  RightHip   FrontHip   LeftHip   RearHip         (4)
+band 2  RightLeg   FrontHip   LeftLeg   RearHip         (transition)
+band 1  RightLeg   LeftLeg                              (2)
+band 0  RightFoot  LeftFoot                             (2)
+```
+
+Slot 0 starts at angle 0 and each spans 45 degrees, so with `atan2(z,x)` the
+mech's **+X is right and +Z is front**. The named faces each cover TWO adjacent
+wedges (Right = slots 7,0 · Front = 1,2 · Left = 3,4 · Rear = 5,6), which means
+**dead ahead is the seam between two Front cells, not the centre of one**.
+
+### Sample cell distributions (Avatar/Mad Cat table A)
+
+```
+band 0  RightFoot   50% rfoot · 30% rdleg · 20% lfoot
+band 1  RightLeg    40% ruleg · 40% rdleg · 10% luleg · 10% ldleg
+band 5  FrontChest  50% utorso · 10% each larm/rarm/ltorso/rtorso/dtorso
+```
+
+Note the scatter is deliberate and generous: a clean foot hit is only half a
+foot hit, and one shot in five crosses to the OTHER foot.
+
+### Playtester-facing visual
+
+An interactive plate of all of the above — every cell of all 8 tables, a plan
+and elevation, and a torso-twist slider that rotates the upper bands live —
+was published for playtesting. Regenerate its dataset with `dmgscan.py`.