Cockpit: punch patches ARE the windows -- all 8 mechs authentic (task #55)

Resolved the non-Madcat blocked views. Rendering every *_COP shell offline
from its mech's AUTHORED eye (SKL [jointeye] translation -- incl. the Thor's
+1.13 offset cockpit) with faces coloured by punch flag showed the answer
plainly: the PUNCH-tagged patches sit exactly over the window apertures in
all 8 mechs. On these texture-less black-diffuse shells dpl_Punchize renders
the tagged geogroup as holes -- the punch patches are the transparent
WINDSHIELDS, and the visible frame is only the non-punch geometry.

The loader now drops the punch batches of _cop meshes (bgfload finish();
BT_COP_KEEPPUNCH=1 keeps them for diagnosis). In-game sweep on pure defaults:
- madcat: the dark dome frame (matches gameplay footage)
- thor / owens / sunder: authentic brace/dash/pillar remnants
- bhk1 / loki / vulture / avatar: authentically FRAMELESS (all-punch
  canopies -- the pod showed no frame for those mechs)

(An early-session test that "disproved" drop-punch ran with the broken
pre-inverse eye; its conclusion was wrong and is corrected in the KB.)

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

context/cockpit-view.md

@@ -87,11 +87,17 @@ inverse(eyeWorld)`. No LookAt anywhere.
   (pos=(0.1,0.15,0.1), neg=(−0.1,−0.15,−0.1) — same for madcat AND bhk1) → the eye equilibrium
   is (0,0,0) and the clamps limit gyro motion to ±0.1..0.15u. The gyro is a BOUNCE/sway
   mechanism, NOT a large steady offset. The authentic resting eye = raw baseOffset for every mech.
-- **bhk1/owens/sunder/vulture (+ loki partial): STILL blocked at rest** — cause now known NOT to
+- **RESOLVED (2026-07-11): the PUNCH patches are the transparent WINDOWS.** Rendering every
-  be the gyro. `BT_EYE_FWD=0.6` still reveals the correct bhk1 view, so the eye-vs-canopy
+  shell OFFLINE from its mech's AUTHORED eye (scratchpad/cop_eyeview.py, faces coloured by punch
-  relationship is off by ~0.6u for that group by some OTHER mechanism (candidates: a joint in the
+  flag) showed the punch panes sitting exactly over the window apertures in all 8 mechs, with the
-  siteeyepoint parent chain we don't pose, per-mech skeleton rest differences, or the canopy
+  visible frame = ONLY the non-punch geometry.  The loader now DROPS the punch batches of `_cop`
-  authored for a view state we don't replicate). OPEN.
+  meshes (bgfload finish(), `BT_COP_KEEPPUNCH=1` keeps them for diag).  In-game sweep on pure
+  defaults, all 8 mechs [T2]: **madcat** = the dark dome frame (matches footage); **thor** =
+  floating brace remnants (authentic; its frame bits are the small non-punch patches); **owens** =
+  dash strip; **sunder** = two side pillars; **bhk1/loki/vulture/avatar** = authentically
+  FRAMELESS (all-punch canopies — the pod showed no frame for those mechs).  NOTE: an
+  early-session in-game test "disproving" drop-punch ran with the broken pre-inverse eye — its
+  conclusion was wrong and cost a detour; with the authentic eye the drop is plainly right.
 - **Hit-bounce: LIVE and verified [T2] (2026-07-11).** The damage→gyro fan-out **FUN_004b2980**
   was re-disassembled byte-exact from the raw image (capstone; scratchpad/dis_4b2980.py) and
   reconstructed as `Gyroscope::ApplyDamageResponse` (gyro.cpp): zero-damage + Collision(0) no-op;

context/open-questions.md

@@ -74,11 +74,11 @@ authentic path scoped.
   inverse view) + single-sided inward-wound lattice + dark `softer` ramp = the dark frame with the
   world through the openings, verified vs gameplay footage; shows by DEFAULT now
   (`BT_HIDE_COCKPIT=1` hides). NOTE the old claim here ("canopy windows are punch texels") was
-  WRONG — the shell has no texture; the openings are geometry. STILL OPEN: the non-Madcat mechs'
+  WRONG — the shell has no texture; the openings are geometry. ✅ RESOLVED (2026-07-11): the
-  views render dark/blocked — NOT an eye-placement bug (disproven: the gyro equilibrium is zero,
+  PUNCH patches are the transparent WINDOWS (dropped by the loader); the frame = the non-punch
-  and the per-mech eye placement is verified EXACT against the SKL jointeye offsets, incl. the
+  geometry only. All 8 mechs render their authentic canopy — madcat's dome, thor/owens/sunder's
-  Thor's authored +1.13 lateral offset cockpit — [[cockpit-view]] §offset-cockpits). It is a
+  brace/dash remnants, and the authentically FRAMELESS bhk1/loki/vulture/avatar. Eye placement
-  canopy-RENDER question (how those shells read from inside). (b) the **3D marker
+  verified exact per-mech incl. Thor's +1.13 offset cockpit ([[cockpit-view]]). (b) the **3D marker
   chain** ([0x31c]/[0x320] dpl objects + the PNAME1-8.bgf player-name meshes over the locked
   target — matters for MP). (c) the compass rotation SIGN is derived-not-verified [T3] (turn
   left → the stem should swing right; flip the sin sign in Draw if footage disagrees).

engine/MUNGA_L4/bgfload.cpp

@@ -1075,6 +1075,32 @@ struct Builder {
     }
 
     void finish() {
+        // COCKPIT WINDSHIELD = THE PUNCH PATCHES (task #55, resolved 2026-07-11): on the
+        // texture-less black-diffuse canopy shells, dpl_Punchize renders the tagged
+        // geogroup as HOLES -- the punch patches are the transparent WINDOWS, and the
+        // visible frame is ONLY the non-punch geometry.  Verified by rendering every
+        // shell from its mech's AUTHORED eye (SKL [jointeye] translation, incl. the
+        // Thor's +1.13 offset cockpit): the punch panes sit exactly over the window
+        // apertures in all 8 mechs; dropping them yields the Madcat's dome frame
+        // (non-punch) and the authentically FRAMELESS views of the all-punch shells
+        // (bhk1/loki/vulture/avatar -- the pod showed no canopy for those).  (An
+        // early-session in-game test that "disproved" drop-punch ran with the broken
+        // pre-inverse eye; its conclusion was wrong.)  BT_COP_KEEPPUNCH=1 keeps the
+        // punch panes (diagnostic).
+        if (meshIsCop) {
+            static int s_keepPunch = -1;
+            if (s_keepPunch < 0) { const char* e = getenv("BT_COP_KEEPPUNCH"); s_keepPunch = (e && e[0] != '0') ? 1 : 0; }
+            if (!s_keepPunch) {
+                std::vector<BgfDrawBatch> kept;
+                kept.reserve(mesh->batches.size());
+                for (const BgfDrawBatch& b : mesh->batches) {
+                    if (b.punch && b.hasRamp && b.texPath.empty())
+                        continue;                       // window pane -> transparent
+                    kept.push_back(b);
+                }
+                mesh->batches.swap(kept);
+            }
+        }
         // COCKPIT CANOPY WINDING ORIENTATION (task #55): the *_cop shells are OPEN strut
         // lattices (38-59% boundary edges, all 12 mechs) drawn SINGLE-SIDED (emitTri) so
         // the openings show the world.  But the authored winding is NOT globally
@@ -1083,9 +1109,7 @@ struct Builder {
         // struts and the other half's BACKS (the solid-box look on BLX/OWX/VUX...).
         // Data-driven fix: orient every canopy face INWARD, toward the mesh interior
         // where the pilot's eye sits, so the whole shell reads correctly from inside
-        // regardless of per-patch mirroring.  (An earlier pass DROPPED the PUNCH-tagged
+        // regardless of per-patch mirroring.  BT_COP_FLIP=1 flips the final orientation
-        // batches believing them the windshield glass -- verified wrong in-game: the
-        // punch patch IS the visible frame.)  BT_COP_FLIP=1 flips the final orientation
         // (diagnostic); BT_COP_DOUBLE=1 disables single-siding entirely (emitTri).
         if (meshIsCop) {
             static int s_flip = -1, s_dbl = -1;