Entry 283

Open questions after the triage: the draw path

render

This entry answers the question left open by VU1 microcode and the model draw path, The port begins: piney-data reads the disc, piney-viewer draws it, Material crop values are a UV animation's reference, not an offset, Animation playback: keys, rotations, loops and morphs, checked against the game, Dungeon rooms join: the room matrix was right, and doors and fog were missing, The viewer plays animations as the game steps them, Morph targets blend in the viewer: cloth waves, How the game draws a field: ground tiles, lighting, cover and object heights, Fields in the viewer, The title's lit icons against VU1: the directional term was doubled, Translucent models showed their hidden layers: FB_ONLY drew colour before Z, ccMorpher::Modify against the game: the weight's lanes and the EE rescale, Mac Anu fixed up: the near plane cut as VU1 cuts it, blades in Kite's hands, the town's lights, and the Chaos Gate, Area 15's story map: EVENTAREA02, the holy ground and the church, its door and lens flare, checked against the game, The window's scaler: sharp bilinear, so thin strokes keep one weight at any size, piros_colour in the field and dungeon: Piros tinted in the fights' draw, Skeith's arena: EVENTAREAB0 for fields 1-8, and how area 27's last door gets there, "The streams' fog", "The field's weather: rain, snow, fireflies and the rest of WORLD::Draw", The animation engine's loose ends: F_Obj, texture offsets, cameras, lights and frame speed, The dungeon doors', the Chaos Gate's and Kite's town footsteps' sounds, The walking PCs' texture variants and steps: changeTEX, rtpcCheckNote, The fog by each vertex's depth, as VU1 writes it, REGION_REPEAT in the GPU renderer: the stream shades' blocks, The dungeons' palette swaps: SetClutList and ChangeClut, The event rooms' direct lights, The field's depth shades, and model Z at the GS's scale, "The dungeon rooms' dressing: SetWater, SetLight, SetObject, SetAnmObject", The game over: ccThGameOver, its noise and the television, The shadow volumes: ccShadowModel, ccShadowPacket and the VU1 programs, The lakes' sky and fireflies: DrawBG and SetRoom's FIREFLYs, The screen shake: cameraShake, cameraShockAbsorber and cameraSet's offset, The Chaos Gate without fog, A party member's skill runs its own animation, The scene file's setup payloads, flag bits and the model header.

FilesUNKNOWNS.md

This entry closes the open questions of 35 entries about the draw path: the VU1 programs and the model packets, animation, fog, lights, shadows, water and the towns', fields' and dungeons' drawing. The triage of 2026-09-28 (UNKNOWNS.md) split every entry's open paragraph into single questions and checked each one against the later log, the docs and the code. The answered ones are listed below with their evidence. What is still open is restated at the end, with the entry that asked it, the play questions first. Open questions of these entries that belong to another subsystem are restated in that subsystem's entry: 282, 284, 285, 286, 287, 288, 289 and 291.

Answered

  • 17: the exact z the V2 unpack delivers — answered by 244 (PCSX2's unpacker copies x, y into z, w; z = S + bits(1/16)); rests on PCSX2's source, not a console
  • 17: the shadow programs beyond their entry points — answered by 160 / docs/engine/shadow.md
  • 26: the unit of the material crop offsets — answered by 29 (a UV animation's reference, not an offset)
  • 26: the lit models' lighting (the viewer's headlight) — answered by 47 (VU1's lighting), 61 (the town's lights)
  • 26: town01's missing water — answered by 28 (wat1, the canal water)
  • 26: animation key interpolation — answered by 32 (keys, rotations, loops and morphs checked)
  • 29: the unit of the run-time UV offset, and where ccSetMaterialPacket makes the ST row — answered by 137 ((u >> 4) & 0xff in 1/256 of the texture, through STROW)
  • 32: how the draw turns ccMaterial.u/v into the ST row — answered by 137
  • 32: the light, camera, ambient, F_Obj and note records — answered by 137 (read where played; F_Obj ported), 139, 278 (notes)
  • 32: whether the field changes ccAnm::frameSpd — answered by 137 (yes: Kite's AnimCtrl, motions, act -5)
  • 32: that the field runs at 30 frames a second — answered by 59 (frame rate 2)
  • 34: SetLight's omni lights and glow effects — answered by 157 (SetLight ported)
  • 34: the clutType 3 and 4 palette swaps — answered by 147 (SetClutList, ChangeClut)
  • 34: GetBG for the lake dungeons, and the white plane in lake doorways — answered by 161 (GetBG picks the sky domes; the sky shows through the doorways)
  • 34: whether the GS fog uses view depth as the viewer assumes — answered by 145 (VU1's per-vertex fogB + fogA * w; piney_draw::DepthFog)
  • 35: the morph blends (F_Morpher cloth) — answered by 36
  • 35, 36: texture-offset animation — answered by 137
  • 35: the animation's transparency channel, read but not drawn — answered by 100 (objects drawn at their own transparency), 137 (F_Obj's localtp)
  • 38: how the fog table's near, far and percentage map to the GS fog — answered by 145 (DepthFog::set_fog is SetFog's arithmetic)
  • 39: the viewer's missing draw window and fade, sky scroll and player-centred background — answered by 72 / crates/piney-world/src/field_area.rs (DrawBG centred on the player, the sky's V scroll), crates/piney-data/src/field/render.rs (FADE_FROM 6700, COVER_RANGE 7200)
  • 39: the viewer's fog depth mapping — answered by 145
  • 47: the desktop's and the field's lit models against VU1 — answered by 47's shared fix, 61 (the town's lights), 141 (tools/test_lights_rs.py: the light group against ccDrawEnv::SetLightMatrix)
  • 47: the "transparency" report — answered by 48 (FB_ONLY drew colour before Z)
  • 61: the merchants, the Recorder, the walking PCs and Orca — answered by 65, 69, 70
  • 61: the gate's menu — answered by 68
  • 61: water, fog, shadows and the arrival's effect in town — answered by 97, 145, 160, 129
  • 61: the user's movement stalls — answered by 62 (a round DualSense stick against a DualShock 2's threshold)
  • 84: the story map's models drawn without SetFog's fog — answered by 145
  • 84: event 11's instructions in the map — answered by 90, 91
  • 107: the condition colours ccChar::Draw applies in the fights — answered by 195 / crates/piney-world/src/foe.rs char_blend (used by the combat draw; foe.rs unit tests)
  • 112: FIREFLY2, the arena's fireflies — answered by 117 / docs/engine/evarea.md (checked against the game's, tools/test_evarea_rs.py)
  • 127: the port's fog per model, not per vertex — answered by 145
  • 130: EntryGimmick in a field — answered by 164
  • 130: WORLD_MAN's two ccBufferSampling depth shades — answered by 152
  • 130: TOBJ's shadow and sounds — answered by 160 (shadow), 150 (the hum)
  • 137: the direct light controller (0x0605) — answered by 151
  • 138: ChangeClut, the door palette swap for clutType 3 and 4 — answered by 147
  • 157: the lakes' fireflies — answered by 161
Still unknown
  • (play) [[38]], [[39]], [[72]]: the field water's run-time textures and scroll, and the clouds' blending, are approximations (docs/engine/field.md, field-walk.md) — every field with water or clouds. - (play) [[97]]: water 0's frame-buffer copy is the whole frame, not the game's 128 x 128 point-sampled one (docs/engine/field-game.md, town02.md) — the towns' water. - (play) [[157]], [[161]]: EntryObject's palettes for the lakes' statues and flowers when GetBG is not 0 are not ported (docs/engine/dungeon.md) — lake dungeons by evening and night. - (play) [[158]]: DUNGEON.fog is not zeroed at the game over's television, and the squeeze moves the whole finished frame, not only what sysLayer's view draws — a game over, in a dungeon most. - (play) [[68]]: the white flash of the gate menu in town is dropped (R::Flash in crates/piney-game/src/world.rs) — leaving a town through the gate. - [[17]]: the DMA order of the matrix, material and model packets (docs/engine/render.md Unknown). - [[17]]: the full bone and skin packet layout, and which of the b/m program families is skin (docs/engine/render.md Unknown). - [[17]]: what view + 0x110 holds, and the mc04b/mc04m variants (docs/engine/render.md Unknown). - [[244]]: the port's ST ignores the 1 + S * 2^-23 factor, 3e-5 of a repeat, below a texel. - [[26]]: textures defined in other files (# entries) show white in the viewer; how # entries resolve is not written down. - [[27]]: the 15 morph-target-shaped models that no F_Morpher in their own file names. - [[36]]: which of the 11,086 F_Morpher records are faces and which cloth; not needed to draw them. - [[55]]: ccMorpher's GetWork first-call path has never run (the same arithmetic). - [[137]]: the spot light controller (0x0607); no Infection room uses one (docs/engine/animation.md Unknown). - [[137]]: F_Obj's flag byte (+0xa2) is not modelled, except its effect-object branch for the streams ([[156]]). - [[137]]: whether an F_Obj pose outlives its clip; F_Obj is not checked in eemu as a whole, only its parts. - [[151]]: whether anything gives the direct light a parent coordinate, and whether the room anm's loop re-applies its frame-0 place. - [[141]]: the areas' light positions and colours at frame 1 are not checked (docs/engine/field-game.md Unknown). - [[141]]: ccDrawEnv +0x88, a second light group SetLightMatrix walks first, is not traced (docs/engine/field-game.md Unknown). - [[61]]: the order in which the gate's light joins the town's; [[141]] checks only the group logic. - [[114]]: how ccLight combines an event room's LGT_ record with SetRoom's place; the block's clump nodes drawn at its matrix (docs/engine/dungeon.md Unknown). - [[160]]: the shadow bounding box check reads vf21, never set; an open edge's second face reads VU1 memory below the directions (game quirks). - [[160]]: shadow modes 2 and 3, never made in Infection, are not ported; the later volumes are not surveyed for them. - [[160]]: EVENTAREA02 and 07 set the light again after GO; noted, not followed. - [[38]]: the field objects' own draw code, and the type-6 FOBJECT2s DrawMesh draws (docs/engine/field.md Unknown). - [[38]]: whether ExtObj-shared object models are lit once or per use. - [[38]]: whether a tile's alpha is clamped past 7,200 (docs/engine/field.md Unknown). - [[38]]: whether VU0 rounds as the modelled truncation; needs a console. - [[61]]: the near-plane rule is read from the VU1 microcode, not run. - [[61]]: the CPU renderer drops triangles behind the camera instead of cutting them (the software model only). - [[48]]: edge rasterisation (the GS's fill rules against the GPU's) and the selection bar's row are not explained. - [[37]], [[48]]: the order of failing and passing pixels where one draw overlaps itself (FB_ONLY's two passes); no test and no known case. - [[37]]: the GS's 8-bit rounding inside the pixel pipeline is not modelled; [[175]] snapped vertex alpha only. - [[48]]: both renderers are ours; neither is compared with a real GS frame capture. - [[53]]: which group, opaque or sorted, a mmat lands in is taken from the disassembly, not compared with the game. - [[57]]: VRAM past the draw buffer reads 0 in the port (docs/engine/stream.md Unknown). - [[121]]: the port's draw clips by the scissor, not by the letterbox's clip box (bboxClipMin/Max). - [[152]]: the GS's bilinear weights and reads past the frame for the depth shades; model Z clamped at 2^32 where the VU saturates. - [[145]]: the per-vertex fog: F values not run against VU1; PRIM.FGE taken from SetFogSw (docs/engine/evarea.md Unknown). - [[157]]: what ccObj::Duplicate copies of the water's object (docs/engine/dungeon.md Unknown). - [[97]]: water 0's place for waterUVModifi2's reach is taken as its root; its coordinate is not checked in a run. - [[84]]: EA_moveTex02's scroll restarts at 0 with a new map after the town, where the game's carries on (docs/engine/evarea.md Unknown). - [[161]]: the lake sky's scroll starts at 0 in each dungeon, where the game's runs on (phase only). - [[99]]: where scFadeDef draws (its layer word +0x94); taken as the font layer. - [[69]]: the fader drawn in the field with the desktop's corner constants; not checked. - [[130]]: the lens flare is left above the picture by the follow camera, and the eye view fails its camera test, as in the game's code; no shot. - [[165]]: the double atan2 is Rust's libm, not newlib's; they agree in every case tried. - [[170]]: whether ccChar::Draw leaves node coordinates stale on a culled frame (docs/engine/battle.md Unknown). - [[177]]: whether the game's GetChunkAdrsF falls back to other loaded files for block 1. - [[264]]: whether EVENTAREA01's DrawObj/DrawObj2 layer differs from objLayer; the port takes the active layer Draw set. - [[26]], [[59]]: whether the game shows town01's floors that dark, and the crisis town's drawing; never compared with a console picture. - [[97]]: what the clouds look like on the console; no capture. - [[99]]: the piros tint's pixels, drawn by the port's fog blend; not compared. - [[96]]: the gate hack's 3D against the game's pixels; the calls are compared. - [[130]]: the balloon's cloth scroll and Ω's runners, not seen in a shot (the story server is Δ). - [[139]]: the walking PCs' variants and steps against the game's pictures and sounds (docs/engine/field-game.md Unknown). - [[145]]: the per-vertex fog against a game picture. - [[146]]: the REGION_REPEAT shades against the game's pictures; the texel choice is checked. - [[147]]: a swapped dungeon against a picture; a twin the file lacks gets a null palette in the game, where the port leaves it unswapped. - [[152]]: the field's depth shades against a game picture. - [[88]]: the sharp-bilinear scaler is not seen at the user's window size, only checked by test; there is no setting for a plain or television-like filter (a port option).