Adds AnimationSequencer.PlayAction as the proper path for Action and
Modifier-class motions (the MotionTable.Modifiers dict, distinct from
Cycles). Action nodes are inserted before the looping cyclic tail so
they drain once and the cycle resumes naturally — leveraging the
sequencer's existing "non-looping head drains, cyclic tail wraps"
queue semantics.
What this does:
- New AnimationSequencer.PlayAction(motionCommand, speedMod=1f):
- Resolves (style<<16) | (motion&0xFFFFFF) from MotionTable.Modifiers
- Falls back to (motion&0xFFFFFF) plain key
- Silent no-op when not found (some motion tables lack these)
- Inserts AnimNodes before _firstCyclic; re-points the cursor when on
the cyclic tail so the action plays immediately
- New MotionCommand.Jump (0x2500003B) + MotionCommand.FallDown (0x10000050)
constants.
- GameWindow.UpdatePlayerAnimation fires PlayAction(Jump) on
result.JumpExtent.HasValue and PlayAction(FallDown) on JustLanded.
Key research finding: retail does NOT animate jumps.
- ACE Player.HandleActionJump explicitly clears PendingMotions and sets
IsAnimating=false during a jump (Player.cs:914-915).
- Empirical verification: the player humanoid's MotionTable only has 8
Modifier entries — all TurnRight/SideStepRight stance variants. No
Jump (0x2500003B) or FallDown (0x10000050) entries.
- Jump is a physics-only action: the character keeps whatever cycle
was active (walk/run/idle) while the physics body arcs through the
air. There is no "raise arms to jump" pose in retail.
PlayAction is still called on jump/land as a safety hatch for creature
Setups that DO carry leap animations in their Modifiers dict (drudge
jumps, monster pounces, etc.). For player humanoids it's a no-op. The
infrastructure is also ready for future emote/combat actions that
legitimately use the Modifiers dict.
470 tests pass, build clean.
Three separate fixes landed today, each addressing a specific bug the
user observed during live play:
1. NPC clothing changes by camera angle (InstancedMeshRenderer)
- Group key was (GfxObjId) only, so every humanoid NPC using the
same body mesh piled into one instance group; only the first
instance's texture was used for the entire DrawInstanced batch,
so which NPC's palette "won" changed as frustum culling and
iteration order shuffled entries.
- Now keyed by (GfxObjId, PaletteHash ^ SurfaceOverridesHash)
so only compatible instances batch; each unique appearance gets
its own draw call. Perf hit is small (humanoid NPCs each emit
one more draw call); visually every NPC is now stable.
2. GpuWorldState dedup on respawn
- Server re-sends CreateObject for the same guid on visibility
refresh / landblock crossing / appearance update. AppendLiveEntity
was blindly appending each time, so GpuWorldState accumulated
multiple copies of the same entity, each with its own
PaletteOverride / MeshRefs. That alone wasn't the clothing bug
(that was #1) but it would have caused other overlap problems
downstream.
- Added RemoveEntityByServerGuid + WorldGameState.RemoveById;
OnLiveEntitySpawnedLocked calls both before creating the new
entity so respawns replace cleanly.
3. Motion wire format — run animation sync with retail observers
- ACE's MovementData constructor only computes interpState.ForwardSpeed
on the WalkForward/WalkBackwards branch; every other ForwardCommand
falls into `else` and passes through WITHOUT speed set, giving
observers speed=0. Sending RunForward directly meant retail
clients saw us "run in place" while position drifted forward.
- Wire: always WalkForward + HoldKey.Run for running. ACE
auto-upgrades to RunForward with creature.GetRunRate() for
broadcast — correct command + correct speed at observers.
- Added per-axis FORWARD_HOLD_KEY / SIDE_STEP_HOLD_KEY /
TURN_HOLD_KEY so every active axis carries HoldKey.Run when
running (matches holtburger's build_motion_state_raw_motion_state).
- Added LocalAnimationCommand to MovementResult so our own
client still plays the RunForward cycle locally while the wire
stays WalkForward. Wire vs. local animation command are now
decoupled.
- Walk-backward wire command changed from WalkForward@-0.65 to
WalkBackward@1.0 (holtburger pattern).
- Strafe speed changed from 0.5 to 1.0 on wire AND local physics
(matches retail sidestep pace).
4. Jump height default + env-var tuning
- Default jumpSkill bumped from 100 → 200 (jump ≈ 3m at full
charge, closer to retail feel for a mid-level character).
- ACDREAM_RUN_SKILL and ACDREAM_JUMP_SKILL env vars now override
the defaults so the user can tune per-character until we parse
PlayerDescription and plumb real skill values through.
5. JustLanded signal on MovementResult
- Tracks airborne→grounded transition so future animation code
can fire the landing cycle when we land. Just a bool flag for
now — no consumer yet (the proper action-queue path will use it).
Not in this commit: jump animation itself. An earlier attempt to
SetCycle(Jump=0x2500003b) fed an Action-type motion into the SubState
cycle resolver, which produced a "torso" mis-render. Reverted. The
proper fix is porting the retail motion action-queue semantics into
AnimationSequencer — see docs/research/deepdives/r03-motion-animation.md
for the spec. That's the next session's work.
470 tests pass, build clean.
Two memory files landed:
1. Updated memory/project_session_2026_04_17.md — covers all three
commits today:
- ff325ab debug overlay + mouse controls
- 7230c15 retail UI research + C# scaffold
- 3f913f1 13-slice deep-dive marathon + scaffolds + roadmap
Includes the "what to build tomorrow" lookup table, the architectural
headline findings, and session lessons (Opus-4.7 parallel swarms are
worth the cost; keystone.dll landmine; GameEvent dispatcher is the
biggest network gap).
2. New memory/project_retail_research_index.md — permanent index for
the 20 research docs (6 UI slices + 13 subsystem slices). Quick-
lookup table "use this slice when you're doing X". Also captures the
critical cross-cutting findings (architecture, wire, dat ranges) and
already-extracted retail-faithful formulas for instant reference.
This file is the standing invariant: before writing any retail-AC-
specific code, open it first to find the matching slice.
Adds the first on-screen HUD for the dev client plus today's mouse-control
refinements. Also lands yesterday's scenery-alignment changes that were
left uncommitted in the working tree.
Overlay:
- BitmapFont rasterizes a system TTF via StbTrueTypeSharp into a 512x512
R8 atlas at startup (Consolas on Windows, DejaVu/Menlo fallbacks)
- TextRenderer batches 2D quads in screen-space with ortho projection;
one shader + two draw calls (rect then text) for panel backgrounds
under glyphs
- DebugOverlay composes info / stats / compass / help panels on top of
the 3D scene; toggles via F1/F4/F5/F6; transient toasts for key events
- DebugLineRenderer and its shaders (carried over from the scenery work)
are properly committed in this commit
Controls:
- Per-mode mouse sensitivity (Chase 0.15, Fly 1.0, Orbit 1.0); F8/F9 to
adjust the active mode multiplicatively (x1.2)
- Hold RMB to free-orbit the chase camera around the player; release
stays at the new angle (no snap-back)
- Mouse-wheel zooms chase distance between 2m and 40m
- Chase pitch widened to [-0.7, 1.4] so mouse-Y tilts both ways from
the default neutral angle
Scenery alignment (carried from yesterday's session):
- ShadowObjectRegistry AllEntriesForDebug + Scale field
- SceneryGenerator uses ACViewer's OnRoad polygon test + baseLoc +
set_heading rotation
- BSPQuery dispatchers accept localToWorld so normals/offsets transform
correctly per part
- TransitionTypes.CylinderCollision rewritten with wall-slide + push-out
- PhysicsDataCache caches visual-mesh AABB for scenery that lacks
physics Setup bounds
The BSP collision detection runs in object-local space, but the
collision response (normals, push offsets) was being applied directly
to world-space SpherePath without rotating back to world space. For
rotated objects (trees, rocks, buildings), this caused the push
direction to be wrong — pushing the player sideways or into the
object instead of away from it.
Added localToWorld quaternion parameter to FindCollisions and all
helper methods (StepSphereDown, CollideWithPt, NegPolyHitDispatch).
All normals and offsets are now transformed via
Vector3.Transform(v, localToWorld) before being applied to SpherePath,
matching ACE's path.LocalSpacePos.LocalToGlobalVec() pattern.
Indoor cell collision uses Quaternion.Identity (cell-local = world).
Object collision passes obj.Rotation.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the patched collision system (~60-70% retail) with a faithful
port of ACE's BSPTree/BSPNode/BSPLeaf/Polygon collision pipeline.
BSPQuery.cs completely rewritten (1808 lines):
- Polygon-level: polygon_hits_sphere_precise (retail two-loop test),
pos_hits_sphere, hits_sphere, walkable_hits_sphere, check_walkable,
adjust_sphere_to_plane, find_crossed_edge, adjust_to_placement_poly
- BSP traversal: sphere_intersects_poly, find_walkable, hits_walkable,
sphere_intersects_solid, sphere_intersects_solid_poly
- BSP tree-level: find_collisions (6-path dispatcher), step_sphere_up,
step_sphere_down, slide_sphere, collide_with_pt, adjust_to_plane,
placement_insert
PhysicsDataCache.cs: Added ResolvedPolygon type with pre-computed
vertex positions and face planes (matching ACE's Polygon constructor
which calls make_plane() at load time). Populated at cache time to
avoid per-collision-test vertex lookups.
TransitionTypes.cs: FindObjCollisions rewritten to use the retail
per-object FindCollisions 6-path dispatcher instead of the old
"find earliest t, then apply custom response" approach. BSP objects
now go through the same collision paths as indoor cell BSP.
The previous approach was explicitly rejected by the user after ~10
iterations of patches. This port follows the CLAUDE.md mandatory
workflow: decompile first → cross-reference ACE → port faithfully.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Session 2026-04-14 summary: rendering rebuild complete, movement
speed+jump+facing working, collision partially working but needs
full retail port.
Memory updated with explicit user feedback: no more patching
collision, must port from decompiled code faithfully per CLAUDE.md.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Document current state (patchwork ~60-70%) and plan for clean port
from ACE's complete C# implementation. Lists all 12 files to port,
what to keep vs replace, and the correct approach.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Indoor CellStruct PhysicsBSP collision for room walls/ceilings.
Dual sphere (body+head) from Setup dimensions.
StepUp attempts before sliding when hitting low obstacles.
FindTimeOfCollision for exact parametric BSP contact time.
Full 6-path BSP dispatcher wired into FindEnvCollisions.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Rewind to t-0.02 instead of exact contact time, plus 2cm normal
push-back. The previous 0.5cm was too small — at high speed the
sub-step could overshoot past the surface.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The CylSphere.Radius from the dat is often smaller than the visual
trunk base. Setup.Radius is the overall bounding radius which better
matches the visual footprint. Use the larger of the two to prevent
clipping into wide tree bases.
Also use Setup.Height as fallback when CylSphere.Height is zero.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Previously CylSphere was only registered when no BSP parts existed.
Retail tests BOTH: CylSphere as the broad collision volume (trunk)
plus BSP parts for polygon-level collision. This ensures the trunk
cylinder catches collisions that individual BSP parts might miss,
especially at the base of large trees.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When FindTransitionalPosition fails (stuck in corner, too many steps),
use the partially-resolved position instead of falling back to the
simple Resolve which has no object collision. This prevents walking
through objects when the transition can't find a clean path.
The player now stops at corners instead of clipping through.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The broad-phase rejection was using 3D distance for cylinder objects,
which includes the Z offset between player feet and cylinder base.
Trees have their origin at the base (Z=ground) while the player
sphere is at chest height (Z=ground+~2.5m). The 3D distance exceeded
the combined radius, causing the collision test to be skipped entirely.
Fix: use horizontal (XY) distance for cylinder broad-phase since
the vertical extent is checked separately in the cylinder test.
Also increase broad-phase margin from 1m to 2m.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
GetNearbyObjects now searches the player's landblock plus all 8
neighbors. Previously only searched one landblock, missing objects
near landblock boundaries — which includes most trees/rocks since
scenery is placed across the full streaming window.
Also added diagnostic logging (will strip after verification).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When collision is detected at t=0 (already overlapping at step start),
push the sphere out along the collision normal by half-radius instead
of trying to slide with zero displacement (which gets stuck).
Returns Adjusted instead of Slid so the transition loop retries
from the pushed-out position.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Restructure FindObjCollisions to compute collision ALONG the movement
path instead of at the final position:
BSP: movement-aware SphereIntersectsPoly with front-face culling
(dot(movement, normal) < 0). Only detects faces the sphere is
approaching, matching retail Polygon.pos_hits_sphere.
Cylinder: quadratic ray-cylinder intersection computes parametric
contact time t. If t < 1.0, sphere is rewound to the contact point.
Both: find the EARLIEST collision (minimum t), rewind sphere to
contact point + small epsilon along normal, then SlideSphere.
This prevents the "walking into walls" penetration (BUG-005).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Closed: jump server packet (002), facing direction (003), run speed (004).
New: collision penetration (005), corner stuck (006), missing trees (007).
All collision bugs stem from static-overlap detection instead of
swept-sphere — needs Transition restructure to use FindTimeOfCollision.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Most scenery objects (trees, rocks) use CylSphere collision from
their Setup, not PhysicsBSP. Register these in ShadowObjectRegistry
with a Cylinder collision type. FindObjCollisions now handles both:
- BSP: full polygon collision via BSPQuery (buildings, stabs)
- Cylinder: radial + vertical cylinder-sphere test (trees, NPCs)
Diagnostics showed 170 CylSphere entities vs 278 BSP entities in
the Holtburg landblock alone — this roughly doubles collision coverage.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace simplified BSP overlap test with retail-faithful 6-path
collision dispatcher. Sphere-intersects-poly now uses movement
vector for front-face culling (prevents wall penetration).
All paths: placement/ethereal, checkWalkable, stepDown, collide,
contact+onWalkable, and default (not in contact).
Ported from ACE BSPTree.cs/BSPNode.cs/BSPLeaf.cs/Polygon.cs,
cross-referenced against decompiled chunk_00530000.c.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace simplified push-out with retail-faithful SlideSphere and
AdjustOffset from transition_pseudocode.md. Crease-projection between
collision normal and contact plane produces smooth wall-sliding.
Object collision uses proper rotation transform to object-local space.
SlideSphere (section 6): computes crease direction via cross product
of collision normal and contact plane normal, projects displacement
onto the crease, then applies the correction offset. Handles three
cases: crease exists, parallel same-direction, parallel opposing.
AdjustOffset (section 6): adds safety check to keep sphere above
contact plane by computing signed distance and pushing up along Z
when the sphere dips below.
FindObjCollisions: removes ad-hoc penetration push-out, now calls
SlideSphere after BSP hit detection for proper wall-slide behavior.
Also fixes: ShadowEntry gains Rotation field, tests updated to match
Register signature, unused variables removed from GameWindow.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Register static entities into terrain cells during streaming.
Transition system queries nearby objects and runs BSP collision.
Player can no longer walk through trees and buildings.
- ShadowObjectRegistry: 24m×24m cell index, Register/Deregister/
RemoveLandblock/GetNearbyObjects matching retail AC's approach
- PhysicsEngine: ShadowObjects property + DataCache wiring point;
RemoveLandblock now also clears shadow objects; TryGetLandblockContext
helper lets Transition resolve landblock id+offset for a world pos
- Transition.FindObjCollisions: queries registry, broad-phase sphere test,
narrow-phase BSPQuery.SphereIntersectsPoly in object-local space,
returns Slid on hit to redirect movement along the surface
- GameWindow.ApplyLoadedTerrainLocked: registers each static entity after
physics BSP data is cached; selects radius from BSP bounding sphere or
Setup.Radius; wires PhysicsDataCache into engine on OnLoad
- 16 new ShadowObjectRegistry unit tests, all 361 tests green
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace simple Z-snap PhysicsEngine.Resolve with ResolveWithTransition
that uses the ported CTransition sphere-sweep pipeline. Movement is
subdivided into sphere-radius steps, terrain collision tested at each
step with step-down for ground contact maintenance.
Falls back to simple Resolve if transition fails. Player controller
now passes pre/post integration positions to the transition system.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
AC heading convention: 0=West, 90=North, 180=East, 270=South.
Our internal yaw: 0=+X (East), PI/2=+Y (North).
Conversion: heading_deg = 180 - yaw_degrees, then holtburger's
from_heading formula: theta=(450-heading).toRad, w=cos(θ/2), z=sin(θ/2).
Previously sent raw Yaw as axis-angle rotation which was ~90° off.
Other clients now see correct facing direction.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Build and send GameAction(Jump) with extent + world-space launch
velocity + sequence counters. Wire format from holtburger
JumpActionData::pack. Server can now validate and replicate jumps
to nearby clients.
Also compute RunRate locally via PlayerWeenie.InqRunRate when
running (server doesn't echo UpdateMotion ForwardSpeed to sender).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two fixes for jump physics:
- Skip ground-snap when velocity Z > 0 (prevents immediate re-landing
at high framerates where per-frame Z delta < 0.05 snap threshold)
- Guard apply_current_movement velocity write behind OnWalkable check
(prevents MotionInterpreter.DoMotion from zeroing jump velocity on
every frame while airborne)
- Guard PlayerMovementController velocity replacement behind OnWalkable
(preserves momentum during airborne flight)
Also fix run speed: compute RunRate locally via PlayerWeenie.InqRunRate
instead of waiting for server UpdateMotion echo (server doesn't echo
to sender). With Run skill 200, run speed is now 9.5 m/s instead of
4.0 m/s.
Strip all diagnostic logging from previous debug sessions.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SmallVelocity threshold (0.25 m/s) in UpdatePhysicsInternal was zeroing
velocity every frame while airborne at the jump apex. With vel~0.01 m/s
and gravity adding only 0.012/frame, the zeroing won every frame and
the character got stuck at peak height forever.
Fix: only apply small-velocity zeroing when OnWalkable (grounded).
While airborne, gravity must accumulate freely through the zero-crossing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two fixes for jump physics:
- Skip ground-snap when velocity Z > 0 (prevents immediate re-landing
at high framerates where per-frame Z delta < 0.05 snap threshold)
- Guard apply_current_movement velocity write behind OnWalkable check
(prevents MotionInterpreter.DoMotion from zeroing jump velocity on
every frame while airborne)
- Guard PlayerMovementController velocity replacement behind OnWalkable
(preserves momentum during airborne flight)
Jump works locally but server packet not yet sent (BUG-002).
Facing direction mismatch logged as BUG-003.
RunRate not verified as BUG-004.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Port FindTransitionalPosition, TransitionalInsert, FindEnvCollisions,
AdjustOffset, DoStepDown, ValidateTransition from transition_pseudocode.md.
Outdoor terrain collision with step-down ground contact. Indoor BSP and
object collision deferred to subsequent tasks.
Also adds PhysicsEngine.SampleTerrainZ() which dispatches the terrain Z
query to the right registered landblock by world-space XY position.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
SpherePath, CollisionInfo, ObjectInfo, TransitionState, PhysicsGlobals.
Types match the pseudocode from transition_pseudocode.md, faithful to
decompiled CTransition + ACE Transition.cs naming.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Load PhysicsBSP and PhysicsPolygons from GfxObj dats during streaming.
BSPQuery.SphereIntersectsPoly traverses the tree for collision detection.
Ported from decompiled FUN_00539270, cross-ref ACE BSPNode.sphere_intersects_poly.
- PhysicsDataCache: thread-safe ConcurrentDictionary-backed cache of GfxObjPhysics
(BSP tree + polygon dict + vertex array) and SetupPhysics (capsule dimensions).
CacheGfxObj/CacheSetup are idempotent — safe to call at every dat load site.
- BSPQuery.SphereIntersectsPoly: recursive BSP descent with bounding-sphere broad
phase, leaf polygon test via existing CollisionPrimitives.SphereIntersectsPoly
(FUN_00539500), and splitting-plane classification for internal nodes.
- GameWindow: _physicsDataCache populated at all GfxObj/Setup dat load sites
(streaming worker path, live-spawn path, ApplyLoadedTerrain render-thread path).
- 6 new unit tests covering null node, bounding-sphere miss, leaf hit, no-contact,
internal node recursion, and empty cache behaviour. All 447 tests green.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Hold spacebar to charge (0→1 over 1s), release to jump. Height from
GetJumpHeight formula using Jump skill via PlayerWeenie. Jump physics
use MotionInterpreter.jump() → LeaveGround() → get_leave_ground_velocity().
JumpExtent is returned in MovementResult (non-null when jump fires this
frame) so GameWindow can log and eventually send the server jump packet.
Double-jump is prevented by jump_is_allowed() checking Contact+OnWalkable
flags before allowing another jump. Tests updated to use charge-then-release
pattern matching the new input model.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implements IWeenieObject with GetRunRate and GetJumpHeight from
decompiled client, cross-referenced against ACE MovementSystem.
Default skills (Run=200, Jump=100) used until skill parsing ships.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Parse ForwardSpeed from UpdateMotion (0xF74C) InterpretedMotionState.
Feed server-echoed RunRate into the player's MotionInterpreter so
get_state_velocity produces the correct speed. Previously hardcoded
at 1.0 (4.0 m/s), now matches character's Run skill.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Four-layer design for retail-faithful movement: speed from RunRate,
charged jump, BSP collision from decompiled CTransition, cell-based
object collision. Decompile-first methodology per CLAUDE.md.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Port ACME's EnvCellManager portal visibility system:
- New CellVisibility class: BFS portal traversal from camera cell,
portal-side clip-plane test, FindCameraCell with grace period
- LoadedCell data populated during streaming (portals, clip planes,
world/inverse transforms, local AABB from CellStruct vertices)
- WorldEntity.ParentCellId tags interior entities for filtering
- InstancedMeshRenderer.Draw accepts optional visibleCellIds set —
interior entities whose parent cell isn't visible are skipped
- Conditional depth clear between terrain and static mesh when
camera is inside a cell (ACME GameScene.cs pattern)
When camera is outdoors, all interiors render (visibleCellIds=null).
When camera enters a building, only BFS-reachable cells render.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Start with BUG-001: wrong cloth textures on characters (observed
during rendering rebuild verification).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace guessed sun direction (0.5, 0.4, 0.6) with ACME's verified
value (0.5, 0.3, -0.3) from GameScene.cs:238. Replace hardcoded
ambient/diffuse (0.25/0.75) with ACME's ambient intensity 0.45 from
LandscapeEditorSettings.cs:108.
Terrain shaders now match ACME Landscape.vert/frag pattern:
- Vertex shader computes Lambert term with xLightDirection uniform
- Fragment shader applies: color * (clamp(lambert, 0, 1) + xAmbient)
Static object shader matches ACME StaticObject.vert:
- LightingFactor = max(dot(N, -L), 0) + ambient
- Removed separate uDiffuseIntensity (ACME doesn't have one)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the single-giant-buffer TerrainRenderer with TerrainChunkRenderer
that groups landblocks into 16×16 chunks, each with its own VAO/VBO/EBO.
Matches ACME's ChunkMetrics/ChunkRenderData/TerrainGPUResourceManager
pattern:
- Pre-allocated max-size VBO (~3.75MB) + EBO per chunk with DynamicDraw
- Incremental glBufferSubData uploads per landblock slot (no full rebuild)
- Fixed slot layout: slot = (localX * 16 + localY), vertBase = slot * 384
- EBO contains only indices for occupied slots → tight draw calls
- One DrawElements per chunk with chunk-level AABB frustum culling
- Empty chunks auto-dispose GPU resources
Streaming-friendly: add/remove a single landblock touches only its slot
in the chunk buffer, not the entire terrain.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace 25 per-landblock VAOs with one shared buffer set. Vertex positions
are now baked in world space during AddLandblock (worldOrigin added to each
vertex), so uModel is eliminated from terrain.vert entirely. Buffer rebuild
happens on the cold path (landblock load/unload) via RebuildGpuBuffers.
Draw loop: bind VAO once, then one glDrawElements per visible landblock
into its sub-range of the shared EBO — same frustum-cull logic, no
VAO/VBO rebind overhead per landblock.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replaces the per-entity glUniform uModel path with a shared instance VBO and
DrawElementsInstanced. All instance model matrices are uploaded to GPU once per
frame; the VAO's per-instance attribute pointers (locations 3–6, divisor=1) are
updated with a byte-offset re-point per group so a single VBO serves all groups
without requiring DrawElementsInstancedBaseInstance (not in Silk.NET 2.23).
Changes:
- InstancedMeshRenderer: add _instanceVbo, _instanceBuffer scratch; EnsureUploaded
sets up mat4 instance attrs (locs 3–6) from the shared VBO; Draw builds the flat
float[] of all instance matrices once then calls DrawElementsInstanced per sub-mesh.
Drops the unused uint TerrainLayer attribute (loc 3 from vertex VBO) — mesh shaders
never used it. Adds InstanceGroup helper to track per-group buffer offsets.
- mesh_instanced.frag: replace sampler2DArray+uTextureLayer with sampler2D uDiffuse,
matching the existing TextureCache / individual-texture pipeline.
- mesh_instanced.vert+frag: track as committed files (were untracked).
- Shader.cs: add SetVec3 helper needed for uLightDirection uniform.
- GameWindow.cs: switch mesh shader load from mesh.vert/.frag to
mesh_instanced.vert/.frag.
Visual output is identical: same entities, same textures, same lighting constants
(SUN_DIR=(0.5,0.4,0.6), AMBIENT=0.25, DIFFUSE=0.75 — moved from frag to vert).
Build: clean. Tests: 431/431 green.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Groups all (entity, meshRef) pairs by GfxObjId before drawing so each
GfxObj's sub-meshes are processed as a contiguous batch. Still uses
per-entity uniform uModel — visual output is identical to the old
StaticMeshRenderer — but the _groups dict is the structural prerequisite
for swapping to DrawElementsInstanced in the follow-up commit.
Key changes:
- New InstancedMeshRenderer.cs with CollectGroups() that fills
_groups[gfxObjId] = List<InstanceEntry> each frame, reusing the
inner List<> objects to avoid per-frame allocation.
- Same two-pass (opaque+clipmap first, translucent second) draw logic
from StaticMeshRenderer, now iterating over groups rather than raw
entity/meshRef pairs.
- GameWindow.cs: field and constructor swapped from StaticMeshRenderer
to InstancedMeshRenderer — public API is identical.
- 431 tests green, 0 warnings.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
TWO root causes for "character disappears when walking far":
1. MarkPersistent stored SERVER GUID but RemoveLandblock checked LOCAL
entity.Id — different namespaces, never matched. Fixed by adding
WorldEntity.ServerGuid field and checking it in RemoveLandblock.
2. Even with rescue working, the player entity stays in its SPAWN
landblock's entity list forever. When the player walks to a new
landblock and the spawn landblock gets frustum-culled, the entity
disappears because neverCullLandblockId is computed from the
player's current position (new landblock) but the entity is stored
in the old landblock.
Fixed by calling GpuWorldState.RelocateEntity every frame in the
player-mode update loop. This moves the entity from whatever
landblock it's currently in to the one matching its actual position.
The scan is O(entities) but only runs for one entity per frame.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
