#91 thor black rectangle: it's the OWN missile pod -- blakskn now material-keyed

Three testers reported a black rectangle swaying with the footsteps in the
thor cockpit (Summoner = the thr1 label, same THX canopy -- no isolation).
Decode: the inside view is a PER-MECH AUTHORED type-A set, not the cop
alone (no fallback in EntitySegment::GetVideoObjectName -- authored data):
madcat/vulture/bhk1 = cop; sunder/loki/avatar = +tor; thor = +tor +MSL
(the shoulder pod, the reported rectangle); owens = +both legs +tshd.
The pod/leg pilot-facing surfaces use the SAME "<pfx>skin:blakskn_dz_*"
interior-structure material as the canopy frame, but the unlit frame
constant was keyed on the _cop FILENAME -- identical material rendered
(0.13,0.12,0.15) on the canopy and pure (0,0,0) on the pod/legs
([matlog]: owx_cop blakskn vcol=FF211F26 vs owx_lule vcol=FF000000).

Fix (bgfload.cpp): the frame-constant treatment keys on meshIsCop OR
material contains "skin:blakskn_dz_". mechfx:blakskn_mtl (tshd shadow
quads) deliberately excluded. Verified: zero pure-black px in the lower
view band across walk captures, the pod plate renders frame-toned and
blends with the bar at rest (the reported anomaly dissolves), owens legs
read as coherent structure, canopy/terrain un-regressed.

Diags added: BT_MAT_LOG=<stem> per-batch material routing dump (bgfload),
BT_HIDE_INSIDE_SEG=<substr> inside-mesh hide (btl4vid), [view] per-segment
inside-roster names. KB: cockpit-view.md exactly-one claim corrected +
the #91 section; bench scratchpad/night11/thorrect.sh.

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

context/cockpit-view.md

@@ -17,8 +17,17 @@ Verified in-game on all 8 mechs (Thor vs gameplay footage). [T2]
 
 - One per mech (12: AVX BLX FIX JAX LOX MAX OWX RAX SNX STX THX VUX; model→skeleton table in
   BTL4.RES @~3232850: `madcat=mad`, `blkhawk=blh`, … — the INSIDE skeleton is the X-variant, so
-  Madcat = `MAX_COP`). It is the torso segment's SkeletonType_A mesh; the inside view loads
-  EXACTLY ONE segment mesh (the `_cop`) — 25 others hidden. [T2]
+  Madcat = `MAX_COP`). It is the torso segment's SkeletonType_A mesh. [T2]
+- **The inside view is a PER-MECH AUTHORED SET, not the cop alone** (#91 decode, 2026-08-04;
+  the earlier "EXACTLY ONE segment mesh" claim was madcat-derived and over-generalized —
+  corrected [T1: no type-A fallback in `EntitySegment::GetVideoObjectName`, SEGMENT.cpp:97 —
+  a segment renders inside ONLY if the resource authors a type-A name; T2: live `[view]`
+  roster logs]). Authored type-A rosters: **madcat / vulture / bhk1 = cop only; sunder /
+  loki / avatar = cop + `<pfx>_tor` torso piece; thor = cop + `thx_tor` + `thx_msl` (the
+  shoulder LRM pod!); owens = cop + BOTH complete legs (`owx_{l,r}{ule,dle,bto,fto}`) +
+  `own_tshd`** — the Owens pilot saw their own legs walking; the Thor pilot saw their
+  missile pod edge bob at the lower frame. `BT_HIDE_INSIDE_SEG=<substr>` hides a non-cop
+  type-A mesh by name (diag, btl4vid.cpp ApplyViewSkeleton).
 - Material: `<pfx>skin:blakskn_dz_{u,r,l}torso_mtl` — DIFFUSE/AMBIENT (0,0,0) + RAMP_REF
   `softer`, **NO texture** (BLXSKIN.BMF is 238 bytes total; byte-verified). The 56 nonzero-UV
   verts in BLX_COP sample nothing (planar-projection authoring leftover). [T1]
@@ -230,6 +239,33 @@ the shell green), `BT_COP_DUMP` (per-batch punch/bbox), `BT_EYE_FWD=<f>` (eye po
 - Feeds: [[gauges-hud]] (the HUD overlays this view) · [[subsystems]] (gyro, task #56)
 - Gotchas: [[reconstruction-gotchas]] §14 (LookAt axis guess), §15 (per-patch edge counting)
 
+## The #91 "black rectangle" (thor) — own-body inside meshes + the blakskn rule [T2]
+
+Three testers (Oracle + RajelAran night 7; Ronin 2026-08-03: "moves diagonally 10→4 with the
+footsteps, at rest sits ON the lower frame bar, right side, half-in") reported a black
+rectangle in the Thor cockpit. Decoded 2026-08-04:
+- The rectangle is the **thor's own `thx_msl` missile-pod mount plate** — authored, authentic
+  own-body geometry (see the roster above; "Summoner" = the `thr1` FE label, same THX canopy,
+  so the cross-reference isolated nothing). It rides the walk-animated shoulder while the eye
+  rides the gyro springs → the diagonal footstep-rhythm sway. Pixel-tracked across walk
+  captures: ~29px wide, x/y oscillating together along the 10↔4 axis, bottom center-right,
+  alternating with the gait [T2, scratchpad/night11/thorrect.sh].
+- **What WAS wrong:** the pilot-facing surfaces of these own-body meshes are authored with the
+  same `<pfx>skin:blakskn_dz_*` interior-structure material as the canopy frame (texture-less,
+  black diffuse, `softer` ramp lo 0.25 → hi 0.99), but the unlit frame-constant treatment was
+  keyed on the `_cop` FILENAME — so the identical material rendered (0.13,0.12,0.15) on the
+  canopy and PURE (0,0,0) on the pod/legs (`[matlog]`-verified: owx_cop blakskn vcol=FF211F26
+  vs owx_lule blakskn vcol=FF000000). Fixed in bgfload.cpp: the constant now keys on
+  `meshIsCop || material contains "skin:blakskn_dz_"`. `mechfx:blakskn_mtl` (tshd shadow
+  quads) deliberately does NOT match. Verified: zero pure-black pixels in the lower view band
+  post-fix, tab renders in the frame tone and blends with the bar at rest; owens legs read as
+  coherent dark structure; canopy/terrain un-regressed. The textured pod/leg batches
+  (THX/GEN/OWEN.BSL pages — bright, mean texel 0.55-0.74) face outboard and were never the
+  visible surfaces.
+- Diags from the dig: `BT_MAT_LOG=<mesh-stem substr>` (bgfload per-batch material routing:
+  resolved colour/texture/ramp/vcol), `BT_HIDE_INSIDE_SEG=<substr>`, `[view]` per-segment
+  inside-roster log. The `softer` ramp resolves lo=(0.25,0.25,0.25) hi=(0.99,0.99,0.99) [T1].
+
 ## Torso-elevation aim -> EYE PITCH (fixed 2026-07-19) [T2, pixel-calibrated]
 User+tester report "pitch does not work": the Torso sim integrated R/F (stick-Y)
 into `currentElevation` (authored limits/rates) but NOTHING consumed it — the

engine/MUNGA_L4/bgfload.cpp

@@ -640,6 +640,7 @@ struct Builder {
     bool currentTSphere = false;   // material is tsphere_mtl (translocation warp): ramp it despite normals
     bool meshIsTSphere = false;    // this OBJECT is the translocation warp -> smooth-tessellate the cone
     bool meshIsCop = false;        // this OBJECT is a *_cop cockpit canopy shell (task #55)
+    std::string meshStem;          // lowercased load stem (BT_MAT_LOG diag)
     std::map<const void*, int> copRoleMap;  // punch-kit roles per PMESH chunk: 1=mask 2=hull 3=skip (see TAG_PATCH)
     int currentCopRole = 0;                 // role of the pmesh being built (consumed by buildPmesh)
     bool currentHasEmissive = false;
@@ -778,7 +779,19 @@ struct Builder {
         // value is not recoverable from the code (texture-less ramp = no texel
         // luminance to index); default (0.13,0.12,0.15) [T3] matches the dark
         // near-black frame in pod gameplay footage.  BT_COP_FRAME="r g b" overrides.
-        if (useRamp && currentTex.empty() && meshIsCop) {
+        // The same interior-structure treatment applies to the blakskn SKIN
+        // family on ANY mesh, not just *_cop shells: the inside skeletons
+        // author pilot-facing surfaces of the OWN-BODY meshes (thor thx_msl
+        // missile pod, owens owx_* legs -- issue #91) with the identical
+        // "<pfx>skin:blakskn_dz_*" material the canopy frame uses, and the
+        // board shaded them all the same way.  Keying on the filename alone
+        // left those surfaces vcol=pure-black -- the reported "black
+        // rectangle" was the thor pod's blakskn mount plate at RGB(0,0,0)
+        // instead of the frame constant.  mechfx:blakskn_mtl (the tshd
+        // shadow quads) must NOT match -- key on the skin-lib "_dz_" form.
+        const bool blaksknSkin =
+            currentMatName.find("skin:blakskn_dz_") != std::string::npos;
+        if (useRamp && currentTex.empty() && (meshIsCop || blaksknSkin)) {
         	auto CB = [](float ff){ int v=(int)(ff*255.0f+0.5f); return (uint32_t)(v<0?0:v>255?255:v); };
         	vcol = 0xFF000000u | (CB(copFrameRGB(0))<<16) | (CB(copFrameRGB(1))<<8) | CB(copFrameRGB(2));
         }
@@ -939,6 +952,28 @@ struct Builder {
             else
                 batch.lodBias = 0.0f;
             mesh->batches.push_back(batch);
+            // #91 diag: BT_MAT_LOG=<substr> dumps each batch of a matching mesh
+            // stem -- the material routing decision (ramp vs lit vs emissive,
+            // resolved colour, texture) for the inside-view black-mesh hunt.
+            {
+                static const char* s_matLog = getenv("BT_MAT_LOG");
+                if (s_matLog && *s_matLog
+                    && meshStem.find(s_matLog) != std::string::npos)
+                    fprintf(stderr, "[matlog] %s mat=%s color=%08X hasDiff=%d "
+                        "tex='%s' ch=%d hasRamp=%d useRamp=%d norms=%d "
+                        "emis=%d vcol=%08X tris=%u\n",
+                        meshStem.c_str(), currentMatName.c_str(), currentColor,
+                        (int)currentHasDiffuse, currentTex.c_str(),
+                        currentTexChannel, (int)currentHasRamp, (int)useRamp,
+                        (int)hasNormals(vtag), (int)pureEmissive, vcol,
+                        idxCount / 3);
+                if (s_matLog && *s_matLog
+                    && meshStem.find(s_matLog) != std::string::npos)
+                    fprintf(stderr, "[matlog]   ramp lo=(%.3f,%.3f,%.3f) "
+                        "hi=(%.3f,%.3f,%.3f)\n",
+                        currentRampLo[0], currentRampLo[1], currentRampLo[2],
+                        currentRampHi[0], currentRampHi[1], currentRampHi[2]);
+            }
             if (getenv("BT_COP_DUMP") && currentHasRamp && currentTex.empty()) {
                 float lo[3]={1e9f,1e9f,1e9f}, hi[3]={-1e9f,-1e9f,-1e9f}, cen[3]={0,0,0}; int nv=0;
                 for (uint32_t i = idxStart; i < idxStart + idxCount; ++i) {
@@ -1386,6 +1421,7 @@ bool LoadBgfFile(const std::string& name, BgfData& out) {
     // single-siding + dark-frame ramp reconstruction in emitTri/buildPmesh applies to
     // these meshes only.
     b.meshIsCop = stemLower(name).find("_cop") != std::string::npos;
+    b.meshStem = stemLower(name);
     b.res = &res;
     for (const Chunk& c : roots) b.collect(c);
     b.finish();

game/reconstructed/btl4vid.cpp

@@ -3193,6 +3193,16 @@ int
 		if (inside && nm != NULL && strstr((const char *)*nm, "_cop") != NULL
 			&& getenv("BT_HIDE_COCKPIT"))
 			nm = NULL;
+		// #91 attribution diag: BT_HIDE_INSIDE_SEG=<substr> hides any OTHER
+		// inside-view (type-A) mesh whose name contains the substring -- the
+		// thor authors THREE type-A meshes (cop + thx_tor + thx_msl) and one
+		// of the extras is the reported black rectangle.
+		{
+			const char *hide = getenv("BT_HIDE_INSIDE_SEG");
+			if (inside && nm != NULL && hide && *hide
+				&& strstr((const char *)*nm, hide) != NULL)
+				nm = NULL;
+		}
 		d3d_OBJECT *obj = NULL;
 		if (nm != NULL)
 		{
@@ -3224,6 +3234,14 @@ int
 		else
 			render_tree.segPick.erase(slot);
 		if (obj) ++shown; else ++hidden;
+		// Inside view: NAME what renders.  The KB says the inside view is the
+		// _cop alone; any OTHER type-A mesh here is a black-material stowaway
+		// (the #91 rectangle investigation) -- keep the roster visible.
+		if (inside && obj != NULL)
+			DEBUG_STREAM << "[view]   shown seg " << slot << ": "
+						 << (const char *)segment->GetName()
+						 << " mesh " << (nm ? (const char *)*nm : "?")
+						 << "\n" << std::flush;
 	}
 	// This reloaded every segment mesh, so every armour-damage binding is stale.
 	BindArmourDamage(viewpoint, render_tree);