Three root causes identified from wowee.log crash at frame 134368:
1. processPendingTransportDoodads() was doing N separate synchronous
GPU uploads (vkQueueSubmit + vkWaitForFences per texture per doodad).
With 30+ doodads × multiple textures, this caused the 489ms stall in
the 'gameobject/transport queues' update stage. Fixed by wrapping the
entire batch in beginUploadBatch()/endUploadBatch() so all texture
layout transitions are submitted in a single async command buffer.
2. Game objects whose M2 model has no geometry/particles (empty or
unsupported format) were retried every frame because loadModel()
returns false without adding to gameObjectDisplayIdModelCache_.
Added gameObjectDisplayIdFailedCache_ to permanently skip these
display IDs after the first failure, stopping the per-frame spam.
3. renderM2Ribbons() only checked ribbonPipeline_ != null, not
ribbonAdditivePipeline_. If additive pipeline creation failed, any
ribbon with additive blending would call vkCmdBindPipeline with
VK_NULL_HANDLE, causing VK_ERROR_DEVICE_LOST on the GPU side.
Extended the early-return guard to cover both ribbon pipelines.
Reads OBJECT_FIELD_SCALE_X (field 4, cross-expansion) from CREATE_OBJECT
update fields and passes it through the full creature and game object spawn
chain: game_handler callbacks → pending spawn structs → async load results
→ createInstance() calls. This gives boss giants, gnomes, children, and
other non-unit-scale NPCs correct visual size, and ensures scaled GOs
(e.g. large treasure chests, oversized plants) render at the server-specified
scale rather than always at 1.0.
- Added OBJECT_FIELD_SCALE_X to UF enum and all expansion update_fields.json
- Added float scale to CreatureSpawnCallback and GameObjectSpawnCallback
- Propagated scale through PendingCreatureSpawn, PreparedCreatureModel,
PendingGameObjectSpawn, PreparedGameObjectWMO
- Used scale in charRenderer/m2Renderer/wmoRenderer createInstance() calls
- Sanity-clamped raw float to [0.01, 100.0] range before use
Extend the locomotion state-change detection to include the WALKING
movement flag. Previously a creature that switched from walking to
running (or vice versa) while staying in the moving state would keep
playing the wrong animation because only the moving/idle transition
was tracked.
Add creatureWasWalking_ alongside creatureWasSwimming_ and
creatureWasFlying_; guard the walking check with isMovingNow to avoid
spurious triggers when the flag flips while the creature is idle.
Clear and erase the new map at world reset and creature/player despawn.
Previously, the animation update for other entities (creatures, players)
was only triggered when the moving/idle state changed. This meant a
creature landing while still moving would stay in FlyForward instead of
switching to Run, and a flying-idle creature touching down would keep
the FlyIdle animation instead of returning to Stand.
Fix: track creatureWasSwimming_ and creatureWasFlying_ alongside
creatureWasMoving_, and fire the animation update whenever any of the
three locomotion flags change. Clean up the new maps on world reset and
on per-creature despawn.
Previously the move-flags callback only tracked SWIMMING and WALKING,
so flying players/mounts always played Run(5) or Stand(0) animations
instead of Fly(61)/FlyIdle(60).
Changes:
- Add creatureFlyingState_ (mirroring creatureSwimmingState_) set by
the FLYING flag (0x01000000) in unitMoveFlagsCallback_.
- Update animation selection: moving+flying → 61 (Fly/FlyForward),
idle+flying → 60 (FlyIdle/hover). Flying takes priority over swim
in the priority chain: fly > swim > walk > run.
- Clear creatureFlyingState_ on world reset.
Add UnitMoveFlagsCallback fired on every MSG_MOVE_* with the raw
movement flags field. Application.cpp uses it to update swimming
and walking state from any packet, not just explicit START_SWIM/
STOP_SWIM opcodes — fixing cold-join cases where a player is already
swimming when we enter the world.
Per-frame animation sync now selects Walk(4) when the WALKING flag is
set, Run(5) otherwise, and Swim(42)/SwimIdle(41) when swimming.
UnitAnimHintCallback is simplified to jump (38=JumpMid) only.
- Add creatureSwimmingState_ map to track which units are swimming
- unitAnimHintCallback with animId=42 (Swim): marks entity as swimming
- unitAnimHintCallback with animId=0 (MSG_MOVE_STOP_SWIM): clears swim state
- Per-frame sync: uses Swim(42)/SwimIdle(41) when swimming, Run(5)/Stand(0) otherwise
— creatures/players now show SwimIdle when standing still in water
- Clear creatureSwimmingState_ on creature/player despawn and world reset
CameraController now transitions the player character to Run (anim 4)
on movement start and back to Stand (anim 0) on stop, guarded by a
prevPlayerMoving_ flag so animation time is not reset every frame.
Death animation (anim 1) is never overridden.
Application creature sync similarly switches creature models to Run (4)
when they move between server positions and Stand (0) when they stop,
with per-guid creatureWasMoving_ tracking to avoid per-frame resets.
processAsyncNpcCompositeResults() had no per-frame budget cap, so when
many NPCs finished async skin compositing simultaneously (e.g. right
after world load), all results were finalized in a single frame causing
up to 284ms frame stalls. Apply the same 2ms budget pattern used by
processAsyncCreatureResults. Load screen still processes all pending
composites without the cap (unlimited=true).
The selection circle was positioned using the entity's game-logic
interpolator (entity->getX/Y/Z), while the actual M2 model is
positioned by CharacterRenderer's independent interpolator (moveInstanceTo).
These two systems can drift apart during movement, causing the circle
to appear under the wrong position relative to the visible model.
Fix: add CharacterRenderer::getInstancePosition / Application::getRenderPositionForGuid
and use the renderer's inst.position for XY (with footZ override for Z)
so the circle always tracks the rendered model exactly. Falls back to
the entity game-logic position when no CharacterRenderer instance exists.
Loading screen now calls processCreatureSpawnQueue(unlimited=true) which
removes the 1-upload-per-frame cap and 2ms time budget, allowing all pending
creature models to upload to GPU in bulk. Also increases concurrent async
background loads from 4 to 16 during load screen. Replaces 40-line inline
duplicate of processAsyncCreatureResults with the shared function.
Move all DBC lookups (CharSections, ItemDisplayInfo), texture path resolution,
and BLP decoding for humanoid NPCs to background threads. Only GPU texture
uploads remain on the main thread via pre-decoded BLP cache.
Move CPU-heavy BLP texture decoding from main thread to background worker
threads for all hot paths: terrain M2 models, WMO doodad M2s, WMO textures,
creature models, and gameobject WMOs. Each renderer (M2, WMO, Character) now
accepts a pre-decoded BLP cache that loadTexture() checks before falling back
to synchronous decode.
Defer WMO normal/height map generation (3 per-pixel passes: luminance, box
blur, Sobel) during terrain streaming finalization — this was the dominant
remaining bottleneck after BLP pre-decoding.
Terrain streaming stalls: 1576ms → 124ms worst case.
- Replace per-frame VMA alloc/free of material UBOs with a ring buffer in
CharacterRenderer (~500 allocations/frame eliminated)
- Batch all ready terrain tiles into a single GPU upload during load screen
(processAllReadyTiles instead of one-at-a-time with individual fence waits)
- Lift per-frame creature/GO spawn budgets during load screen warmup phase
- Add background world preloader: saves last world position to disk, pre-warms
AssetManager file cache with ADT files starting at app init (login screen)
so terrain workers get instant cache hits when Enter World is clicked
- Distance-filter expensive collision guard to 8-unit melee range
- Merge 3 CharacterRenderer update loops into single pass
- Time-budget instrumentation for slow update stages (>3ms threshold)
- Count-based async creature model upload budget (max 3/frame in-game)
- 1-per-frame game object spawn + per-doodad time budget for transport loading
- Use deque for creature spawn queue to avoid O(n) front-erase
- Async creature model loading: M2 file I/O and parsing on background threads
via std::async, GPU upload on main thread when ready (MAX_ASYNC_CREATURE_LOADS=4)
- CharSections.dbc lookup cache: O(1) hash lookup instead of O(N) full DBC scan
per humanoid NPC spawn (was scanning thousands of records twice per spawn)
- Frame time budget: 4ms cap on creature spawn processing per frame
- Wolf/worg model name check cached per modelId (was doing tolower+find per
hostile creature per frame)
- Weapon attach throttle: max 2 per 1s tick (was attempting all unweaponized NPCs)
- Separate texture application tracking (displayIdTexturesApplied_) so async-loaded
models still get skin/equipment textures applied correctly
Reset descriptor pools in CharacterRenderer/M2Renderer/WMORenderer on map
change to prevent VK_ERROR_DEVICE_LOST from pool exhaustion. Defer re-entrant
SMSG_NEW_WORLD during active world load to avoid recursive cleanup crashes.
Gate swim bubbles on swimming state, skip redundant shadow pipeline re-init,
add WOWEE_SKIP_* env vars for render isolation debugging.
- add one-time warning caches for missing CreatureDisplayInfo and missing model path lookups
- log each missing displayId/model-path only once instead of every spawn retry/frame
- remove duplicate empty-model-path warning in spawnOnlineCreature (already reported by lookup path)
This cuts high-frequency log I/O and string formatting in hotspots when server sends unknown displayIds, while preserving first-occurrence diagnostics.
- resolve creature virtual item values as Item.dbc entry -> DisplayID before direct ItemDisplayInfo fallback\n- use correct WotLK virtual item base index set (starting at 56) and keep retry attach loop for late update fields\n- keep NPC main-hand weapon attachment path and clear erroneous offhand attachments\n- refactor NPC virtual weapon attach into helper with per-creature retry tracking\n- tighten humanoid facial hair geoset selection to avoid loading all beard/sideburn meshes\n- adjust facial 3xx preference to favor chin-hair channel when available
Hide NPC cloak/object-skin mesh when no cape texture resolves by using a transparent texture fallback, preventing skin-texture bleed on cloaks. Tighten NPC equipment region compositing by slot and add safe humanoid geoset selection to avoid robe-over-pants conflicts and odd pants texturing.
Reduce login/runtime hitching by deferring non-critical world-system initialization across frames, lowering per-frame transport doodad spawn budget, and demoting high-volume transport/MO_TRANSPORT diagnostics to debug. Gate M2 glow diagnostics behind WOWEE_M2_GLOW_DIAG and make zone music prewarm opt-in via WOWEE_PREWARM_ZONE_MUSIC.
- Add per-frame cap for first-time creature model loads to spread expensive model/texture decode work across frames instead of burst loading.
- Keep cached creature spawns flowing while deferring uncached display IDs, preserving world population updates with smoother frame pacing.
- Defer transport WMO doodad instancing into a queued batch processor with a strict per-frame budget to avoid multi-hundred-ms spikes when ships/elevators register.
- Process deferred transport doodad batches in both normal gameplay update and loading-screen warmup loop so heavy transport setup can be amortized before and after world entry.
- Cleanup pending transport doodad batches on gameobject despawn to prevent stale work and avoid attaching children to removed parents.
- Lower character texture cache miss logging from INFO to DEBUG to reduce log I/O contention during movement-heavy asset streaming.
- keep character-selection state until world entry load/finalize completes
- move expensive post-load setup (test transport + creature callback prep) before loading screen shutdown
- add bounded world warmup pass under loading screen to drain initial network/spawn backlog
- start intro camera pan after warmup so rotation begins when gameplay becomes visible
- guard test transport setup so it runs once per session
- add per-update world socket parse budget to prevent single-frame packet-drain stalls
This reduces visible 3-4s stutter after login by shifting startup work behind the loading screen and time-slicing packet processing.
- Use interpolated moveInstanceTo() for normal server NPC position deltas so walk/run animations play instead of visual sliding
- Keep hard snaps only for dead units and large corrections/teleports
- Add per-creature render position cache to drive interpolation safely across frames and clear it on spawn/despawn/reset
- Keep minimap controls visible regardless of quest-status availability
- Correct minimap NPC/quest marker projection mapping to match minimap shader transform
- Add optional nearby NPC minimap dots setting (default OFF), exposed in Gameplay > Interface and persisted in settings
Selection ring rendering has been upgraded to better match WoW-like target feedback and remain readable across complex geometry.
Visual updates:
- Reworked ring mesh/shader from a simple two-band strip to a unit-disc + radial fragment shaping.
- Implemented a thinner outer ring profile.
- Added an inward color/alpha gradient that fades from the ring toward the center.
Placement/anchoring updates:
- Added CharacterRenderer::getInstanceFootZ() to query model foot plane from instance bounds.
- Added Application::getRenderFootZForGuid() to resolve per-GUID foot height via live instance mapping.
- Updated GameScreen target selection placement to anchor the effect at target foot Z.
Ground/surface stability:
- In renderSelectionCircle(), added floor clamping against terrain, WMO, and M2 floor probes at target XY.
- Raised final placement offset to reduce residual clipping on uneven surfaces.
Depth/visibility behavior:
- Added polygon offset during ring draw to reduce z-fighting.
- Disabled depth testing for the selection effect draw pass (with state restore) so the ring remains visible through terrain/WMO occluders, per requested behavior.
State safety:
- Restored modified GL state after selection pass (depth test / polygon offset / depth mask / cull).
Build validation:
- Verified with cmake build target "wowee" after each stage; final build succeeds.
Classic: synchronized Data/expansions/classic/opcodes.json to /home/k/Desktop/classicopcodes.h with exact symbol/value parity (0 missing, 0 mismatches).
WotLK: synchronized Data/expansions/wotlk/opcodes.json to /home/k/Desktop/azerothcoreOpcodes.h and aligned symbol names to AzerothCore naming.
Logical opcode layer: expanded include/game/opcode_table.hpp and src/game/opcode_table.cpp to include missing canonical opcode symbols required by synced tables, and removed legacy alias fallback block so canonical names are used directly.
Gameplay/handler updates included from ongoing fixes: duel/taxi stale opcode cleanup, level-up/sound handling adjustments, and related parser/packet references updated to match canonical opcode identifiers.
Validated by successful full build: cmake --build build -j32.
Two bugs caused the client to look like a bot to server GMs:
1. Strafe animation played during forward+strafe (W+A) instead of the
walk/run animation. Added pureStrafe guard so strafe animations only
play when exclusively strafing (no forward key or auto-run active).
2. CMSG_MOVE_SET_FACING was never sent on mouse-look turns. The server
predicts movement from the last known facing; without SET_FACING the
heartbeat position appeared to teleport each time the player changed
direction. Now sent at up to 10 Hz whenever facing changes >3°,
skipped while keyboard-turning (handled server-side by TURN flags).
- Clear introActive/idleOrbit in externalFollow block so mouse panning works during taxi
- Skip full world reload on same-map teleports (taxi landing) by tracking loadedMapId
- Collect all model IDs for a path when resolving gryphon display ID (fixes displayId=0)
- Remove incorrect MountDisplayId fields from Vanilla/Turtle TaxiNodes DBC layouts
- Add Vanilla fly animation IDs (234/229/233) to taxi mount animation candidates
Defer equipment texture compositing to a queue processed max 1 per frame
instead of compositing immediately on callback (each compositeWithRegions
call does file I/O + CPU blit + GPU upload on the main thread). Reduce
MAX_SPAWNS_PER_FRAME from 96 to 8 and increase inspect rate limit from
0.75s to 2s. Demote noisy per-frame logs to DEBUG level.
Remove HDPackManager, expansion overlay manifests, and BLP size-comparison
logic. Assets now resolve through a single manifest with a simple override
directory (Data/override/) for future HD upgrades.
Expansion overlays allow each expansion to supplement the base asset data
via an assetManifest field in expansion.json, loaded at priority 50 (below
HD packs). The asset extractor gains --reference-manifest for delta-only
extraction. Also fixes CharSections field indices (VariationIndex=4,
ColorIndex=5, Texture1=6) across all DBC layout references.
- Vanilla M2 bone struct (108 bytes) with 28-byte animation tracks
- Version-aware bone parsing (vanilla vs WotLK format detection)
- Fix CharSections.dbc field layout for vanilla (variation/color at 4-5)
- Remove broken CharSections.csv files (all fields marked as strings)
- Expansion data reload on profile switch (DBC cache clear, layout reload)
- Vanilla packet encryption (VanillaCrypt XOR-based header crypt)
- Extended character preview geoset range (0-99) for vanilla models
- DBC cache clear support in AssetManager
Replace hardcoded WotLK protocol constants with a data-driven architecture
supporting Classic 1.12.1, TBC 2.4.3, and WotLK 3.3.5a. Each expansion
has JSON profiles for opcodes, update fields, and DBC layouts, plus C++
polymorphic packet parsers for binary format differences (movement flags,
speed fields, transport data, spline format, char enum layout).
Key components:
- ExpansionRegistry: scans Data/expansions/*/expansion.json at startup
- OpcodeTable: logical enum <-> wire values loaded from JSON
- UpdateFieldTable: field indices loaded from JSON per expansion
- DBCLayout: schema-driven DBC field lookups replacing magic numbers
- PacketParsers: WotLK/TBC/Classic parsers with correct flag positions
- Multi-manifest AssetManager: layered manifests with priority ordering
- HDPackManager: overlay texture packs with expansion compatibility
- Auth screen expansion picker replacing hardcoded version dropdown
Transport System (Phases 1-7):
- Implement TransportManager with Catmull-Rom spline path interpolation
- Add WMO dynamic transforms for moving transport instances
- Implement player attachment via world position composition
- Add test transport with circular path around Stormwind harbor
- Add /transport board and /transport leave console commands
- Reuse taxi flight spline system and external follow camera mode
NPC Spawn Fixes:
- Add smart ocean spawn filter: blocks land creatures at high altitude over water (Z>50)
- Allow legitimate water creatures at sea level (Z≤50) to spawn correctly
- Fixes Elder Grey Bears, Highland Striders, and Plainscreepers spawning over ocean
- Snap online creatures to terrain height when valid ground exists
NpcManager Removal:
- Remove deprecated NpcManager (offline mode no longer supported)
- Delete npc_manager.hpp and npc_manager.cpp
- Simplify NPC animation callbacks to use only creatureInstances_ map
- Move NPC callbacks to game initialization in application.cpp
Water Rendering:
- Fix tile seam gaps caused by per-vertex wave randomization
- Add distance-based blending: seamless waves up close (<150u), grid effect far away (>400u)
- Smooth transition between seamless and grid modes (150-400 unit range)
- Preserves aesthetic grid pattern at horizon while eliminating gaps when swimming
Replace 2D ImGui text markers with proper 3D billboard sprites using BLP textures.
Features:
- Billboard rendering using Interface\GossipFrame\ BLP textures (yellow !, yellow ?, grey ?)
- WoW-style visual effects: bob animation, distance-based scaling, glow pass, distance fade
- Proper NPC height positioning with bounding box detection
- Camera-facing quads with depth testing but no depth write
- Shader-based alpha modulation for glow and fade effects
Technical changes:
- Created QuestMarkerRenderer class with billboard sprite system
- Integrated into Renderer initialization for both online and offline terrain loading
- Rewrote updateQuestMarkers() to use billboard system instead of M2 models
- Disabled old 2D ImGui renderQuestMarkers() in game_screen.cpp
- Added debug logging for initialization and marker tracking
Quest markers now render with proper WoW visual fidelity.
Loads and renders actual quest marker M2 models from
World\Generic\PassiveDoodads\Quest\ as floating 3D objects above NPCs
based on quest status, replacing 2D ImGui text markers.
Features:
- Loads QuestExclamation.m2 (yellow !) for available quests
- Loads QuestQuestionMark.m2 (silver ?) for completable quests
- Updates marker positions dynamically as NPCs move
- Automatically spawns/despawns markers based on quest status changes
- Positions markers above NPC heads using render bounds
Quest markers are now proper 3D assets consistent with WoW 3.3.5a client.
After creating a character, automatically select it in the character
list instead of defaulting to the previously selected character.
Changes:
- Added selectCharacterByName() method to CharacterScreen
- Store created character name in Application
- On creation success, auto-select the new character by name
- Falls back to saved selection if new character name doesn't match
