#124 CORRECTION: the frame adapter is a Z-REFLECTION, not a pi rotation --

proven by the mech's own pods

The first fix (c5dfb00) negated x AND z. The operator challenged it:
'they couldn't hit rear panels AT ALL -- a 180 rotation couldn't explain
that' -- and a rotation would also have relocated rear damage, not
erased it, while the bench premise (the dummy faces its attacker) was
assumed, never measured.

The zero-premise probe (BT_ASPECT_TEST): four self-impacts at known
WORLD cardinals + one at each weapon's PHYSICAL MUZZLE position -- the
mech's own asymmetric geometry as the left/right anchor. Measured:

* WorldToLocal is a clean R(yaw), no hidden terms (yaws -30/150/40 deg
  all consistent);
* our frame: RIGHT = +X -- SAME as the 1995 ring (W0/W7 = Right*) --
  and FORWARD = -Z, flipped vs the ring's Front*-in-+Z-arc;
* under the pi rotation the pods CROSSED (left muzzle -> rtorso);
  under z-negation every anchor lands its own side.

So: negate local z only, and -- since a reflection reverses angular
direction -- the SelectSlice torso-twist term flips sign with it
(twist was 0 in all probes; twisted-torso verify tracked on #124).

Anchor acceptance (fresh exe, third yaw): LRM15_1 left pod -> ltorso,
LRM15_2 -> right wedges, left/right arm muzzles -> larm/rarm, nose ->
front cells, tail -> rear cells, flank cardinals -> rgun / left wedges.

The BT_ASPECT_TEST probe + BTWeaponMuzzleWorld bridge stay as bench
scalpels.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Joe DiPrima
2026-08-03 09:05:35 -05:00
co-authored by Claude Fable 5
parent c5dfb002f3
commit b0b7adc4ea
4 changed files with 114 additions and 24 deletions
+15 -9
View File
@@ -1012,15 +1012,21 @@ code involved; grammar = the binary ctors @0x49ea48/@0x49e5e4). Results:
"Right*", W1/W2 = "Front*" -- with theta = atan2(z,x) from +X toward +Z, the
mech-local FRONT is the **+Z arc (theta 45-135 deg)** and dead-ahead is the
W1|W2 SEAM. (Corrects the earlier #92 comment that put frontal hits in W6.)
* **⚠ PORT FRAME ADAPTER (#124, fixed 2026-08-03)**: that +Z-forward is the
**1995 resolver frame**. OUR engine's entity frame carries forward along
**Z** (bench-measured: a victim facing its attacker took frontal beams at
local z=4..10, θ≈270°, resolving the REAR-named cells — which was the whole
night-10 field table: rear panels unhittable, Owens flanks CROSSED). The
frames differ by a **π rotation about Y** (rotation, not reflection — the
flank observations cross rather than persist), so `ResolveHit` negates
local x and z after `WorldToLocal` before the band/wedge math. A/B verified:
frontal fire θ≈1.5 → utorso/dtorso/ltorso/rtorso, zero rear zones. The
* **⚠ PORT FRAME ADAPTER (#124, fixed 2026-08-03; corrected same day by the
MUZZLE-ANCHOR probe)**: that +Z-forward (and +X-right, from the W0/W7
"Right*" names) is the **1995 resolver frame**. OUR engine frame, measured
with the mech's own asymmetric geometry as the anchor (`BT_ASPECT_TEST`:
four world-cardinal self-impacts + one at each weapon's physical muzzle):
**RIGHT = +X, same as 1995; FORWARD = Z, flipped** — the LEFT missile pod
sits at raw local x=2.32, the RIGHT at +2.32, and `WorldToLocal` is a clean
R(yaw) with no hidden terms (verified at yaws 30°/150°/40°). The frames
differ by a **Z-NEGATION ONLY — a reflection** (the first commit's π
rotation CROSSED the pods: the left muzzle resolved rtorso). `ResolveHit`
negates local z; and because a reflection reverses angular direction, the
torso-twist term in `SelectSlice` enters with the OPPOSITE sign (twist was 0
in every probe — the twisted-torso verify is the tracked follow-up).
Anchor acceptance: left pod→ltorso, right pod→right wedges, left/right arm
muzzles→larm/rarm, nose→front cells, tail→rear cells, at a third yaw. The
separate "one band LOW" Y-signature is #16 (pick-ray parallax), not this.
Frontal unaimed foot-band hits therefore straddle the RightFoot/LeftFoot
cells by impact-x sign, and the authored cross-bleed (20% opposite foot,
+21 -15
View File
@@ -177,7 +177,11 @@ DamageZonePercentTable *
if (rotateWithTorso) // binary this[3] != 0
{
theta += owner->TorsoHeading(); // torso+0x1d8
// #124 frame adapter, angular half: theta arrives in the 1995 frame
// (ResolveHit's z-reflection), so the twist -- measured in OUR frame
// -- enters with the OPPOSITE sign (a reflection reverses angular
// direction). Binary form: theta += torso+0x1d8.
theta -= owner->TorsoHeading(); // torso+0x1d8 (sign: see adapter)
if (theta >= TwoPi) theta -= TwoPi;
if (theta < 0.0f) theta += TwoPi;
}
@@ -277,20 +281,22 @@ int
Scalar heightRef = owner->CylinderReferenceHeight(); // owner+0x2ec[+0xc]
if (heightRef <= 0.0f) return -1;
// PORT FRAME ADAPTER (#124, measured 2026-08-03). The 1995 resolver frame
// carries the mech's FORWARD along +Z: the authored wedge ring puts the
// FrontChest-family cells in the +Z arc (slice-name audit [T1]) -- frontal
// hits MUST read z > 0 for the shipped tables to mean what they say. Our
// engine's entity frame carries forward along -Z (bench 2026-08-03: the
// dummy facing its attacker takes frontal beams at local z = -4..-10,
// theta ~270 deg, resolving the REAR-named cells; the night-10 field table
// -- rear panels unhittable, Owens left torso -> RIGHT pod -- is this same
// pi ROTATION at every aspect: rotation, not reflection, because the
// left/right flanks CROSS rather than persist). Express the impact in the
// cylinder's frame: rotate pi about Y. y (the band axis) is untouched,
// and the frames share handedness, so the torso-twist add in SelectSlice
// keeps its sign.
local.x = -local.x;
// PORT FRAME ADAPTER (#124, corrected 2026-08-03 by the MUZZLE-ANCHOR
// probe). The 1995 resolver frame: FORWARD = +Z (the authored ring puts
// Front* cells in the +Z arc) and RIGHT = +X (W0/W7 = "Right*") [T1
// slice-name data]. Our engine frame, measured with the mech's own
// asymmetric geometry as the anchor (BT_ASPECT_TEST muzzle probes -- the
// LEFT missile pod LRM15_1 sits at raw local x=-2.32, the RIGHT pod
// LRM15_2 at x=+2.32): RIGHT = +X (SAME as 1995) but FORWARD = -Z
// (drive-code convention, "forward = -Z at heading 0"). The frames
// therefore differ by a Z-NEGATION ONLY -- a REFLECTION, not the pi
// rotation first committed (that crossed the pods: left muzzle resolved
// rtorso). y (the band axis) untouched.
//
// A reflection reverses the angular walk direction, so the OTHER angular
// input -- the torso-twist term SelectSlice adds -- must flip sign with
// it (applied there, same adapter). Twist was 0 in every probe; the
// twisted-torso verify is the follow-up on #124.
local.z = -local.z;
// --- height -> layer (penetration depth) ---
+68
View File
@@ -6218,6 +6218,74 @@ void
}
}
// #124 PROBE (BT_ASPECT_TEST=1): the zero-premise frame probe. At frame
// ~700, dispatch four self TakeDamage messages whose impact points sit
// at KNOWN WORLD BEARINGS (+X/-X/+Z/-Z, 8 u out, torso height) around
// our OWN mech, logging our yaw + each resolve. With the mech's
// heading known and torso twist zero, the theta each bearing produces
// measures the WorldToLocal frame convention directly -- no facing
// premises, no shooter geometry.
if ((Entity *)this == application->GetViewpointEntity()
&& getenv("BT_ASPECT_TEST"))
{
static int s_apFrame = 0;
if (++s_apFrame == 700)
{
YawPitchRoll ypr; ypr = localOrigin.angularPosition;
DEBUG_STREAM << "[aspect] mech yaw=" << (float)ypr.yaw
<< " twist=" << (float)TorsoHeading()
<< " pos=(" << localOrigin.linearPosition.x << ","
<< localOrigin.linearPosition.z << ")\n" << std::flush;
static const struct { const char *tag; float dx, dz; } kProbe[4] =
{ {"+X", 8, 0}, {"-X", -8, 0}, {"+Z", 0, 8}, {"-Z", 0, -8} };
for (int pi = 0; pi < 4; ++pi)
{
Damage dmg;
dmg.damageType = (Enumeration)3; // Laser
dmg.damageAmount = 0.01f; // negligible
dmg.burstCount = 1;
dmg.impactPoint = localOrigin.linearPosition;
dmg.impactPoint.x += kProbe[pi].dx;
dmg.impactPoint.y += 7.0f; // torso height
dmg.impactPoint.z += kProbe[pi].dz;
DEBUG_STREAM << "[aspect] probe " << kProbe[pi].tag
<< " ->\n" << std::flush;
Entity::TakeDamageMessage td(
Entity::TakeDamageMessageID, sizeof(Entity::TakeDamageMessage),
GetEntityID(), -1, dmg, -1);
Dispatch(&td);
}
// SELF-ANCHORED L/R probe: impact each weapon's PHYSICAL muzzle
// position -- the pod's own asymmetric geometry is the anchor.
// A right-side launcher's muzzle impact must resolve a
// right-family zone; which flip (z-only reflection vs full pi
// rotation) achieves that decides the adapter's x term.
{
extern int BTWeaponMuzzleWorld(Subsystem *w, Point3D *out); // mechweap TU
for (int wi = 2; wi < GetSubsystemCount(); ++wi)
{
Subsystem *w = GetSubsystem(wi);
Point3D mzp;
if (w == 0 || !BTWeaponMuzzleWorld(w, &mzp))
continue;
Damage dmg;
dmg.damageType = (Enumeration)3;
dmg.damageAmount = 0.01f;
dmg.burstCount = 1;
dmg.impactPoint = mzp;
DEBUG_STREAM << "[aspect] muzzle probe '"
<< (w->GetName() ? w->GetName() : "?")
<< "' at (" << mzp.x << "," << mzp.y << "," << mzp.z
<< ") ->\n" << std::flush;
Entity::TakeDamageMessage td(
Entity::TakeDamageMessageID, sizeof(Entity::TakeDamageMessage),
GetEntityID(), -1, dmg, -1);
Dispatch(&td);
}
}
}
}
// #86 BENCH (BT_KILL_SUBSYS=<name>): force the AUTHENTIC destruction of
// one named subsystem -- MechSubsystem::ForceCriticalFailure, the exact
// call the zone crit-cascade (SendSubsystemDamage) makes, and the thing
+10
View File
@@ -997,3 +997,13 @@ void BTWeaponSetViewFireEnable(Subsystem *sub, int enable)
if (sub != 0 && sub->IsDerivedFrom(MechWeapon::ClassDerivations))
((MechWeapon *)sub)->SetViewFireEnable(enable);
}
// #124 aspect probe: a weapon's physical muzzle WORLD position (the pod's own
// asymmetric geometry as a left/right anchor). -1-family returns 0.
int BTWeaponMuzzleWorld(Subsystem *sub, Point3D *out)
{
if (sub == 0 || out == 0 || !sub->IsDerivedFrom(MechWeapon::ClassDerivations))
return 0;
((MechWeapon *)sub)->MuzzlePoint(*out); // the public beam-render alias
return 1;
}