48d9506cf15d8178d0e5d526d5dc67f24e2ee155
6
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 ReaperOfVeriod
|
48d9506cf1 |
Decode tfrag texture bindings, vertex colours and light records
Adds level_core.rs (the uncompressed core index header and texture table) and extends tfrag.rs with the data a draw call needs beyond geometry. An ad-gif is five A+D quadwords holding packed fields rather than GS registers; the level load rewrites them in place before drawing. On disc TEX0_1.data_lo indexes the level's texture table, TEX1_1.data_lo is the LOD bias K and data_hi the minification filter, and CLAMP_1 holds both wrap modes. Three things are easy to get wrong here: * K is a signed 12-bit value in 1/16 units, sign-extended by shifting the field up 4 and back down. Retail values are -137..-89 mip-scaled; reading it unsigned gives 1911 instead of -137. * TEX1_1.data_hi is MMIN, and 4 decodes as LinearMipmapNearest, not LinearMipmapLinear as the raw bit layout suggests. * RGBA and the light/normal records are not in the VIF lists at all. They sit at fixed offsets in the patch data block: rgba_size*4 colours at rgba_ofs, vert_count light records at light_ofs+0x10. The tfrag origin lives there too -- if no STROW row supplied one, the quadword at light_ofs is it. The texture table checks out on its own terms: 93 entries at index offset 0x2900 on level 0, and 0x2900 + 93*16 lands exactly on the next range. Verification: an ad-gif index outside the level's table would mean the geometry and the textures were misread independently, so every one is range-checked against real entries. All 27 levels: 47,950 ad-gifs over 18,102 patches, all in range, referencing 1429 of 1595 level texture entries (90%). Each patch's light array is exactly vert_count records and its RGBA array exactly rgba_size*4. Still open: LOD selection and morphing, and the texel data itself. |
||
|
ReaperOfVeriod
|
9459f15329 |
Parse tfrag topology: vertex info, strips, indices and triangles
Turns tfrag positions into drawable geometry. Every unpack in a patch's VIF lists is now classified by format *and* destination VU address, using the VU header at the head of the common list. That table is what tells the strips from the indices from the parent arrays: they are all V4_8 and only the address distinguishes them. Strip indices address the vertex-info array, not the position array. Several vertex-info entries can share a position, so an index resolves through vertex_info[i].vertex / 2 to reach a position. Vertex-info also carries UVs and the LOD parent link. A strip run of n vertices is a GS triangle strip of n - 2 triangles with alternating winding. The vertex count is biased by -128 when negative, and the first record of a list is always biased even when positive -- real patches open with -116 or -124, never a small positive number. tri_count lives in the patch header and cannot be derived from the lists without walking them, so reproducing it is a strong check: all 27 levels, 18,102 patches, 809,879 triangles, every patch matching its header, every triangle index inside its LOD's vertex-info range. Four wrong turns got there, each of which failed quietly rather than loudly: reading the element size from the immediate instead of the command, 2-byte payload padding instead of 4, ignoring the inline payload of STROW/STCOL/STMASK/MPG/DIRECT, and banking strips by arrival order instead of by LOD. Every one produced parseable, plausible geometry. Only the header's own triangle count caught them. Still open: ad-gif texture bindings, per-vertex RGBA and light records, and the LOD switch and morph rules. |
||
|
ReaperOfVeriod
|
36641b4cf8 |
Parse tfrag terrain blocks and decode all vertex positions
Adds gc-formats::tfrag and a gated real-disc test that decodes every terrain patch on all 27 levels. Two rules decide whether this works at all, and both were wrong on the first attempt: The vector format is encoded in the VIF command byte, not the immediate. Bits 2-3 are vn (component count minus one) and bits 0-1 are vl, giving element_size = ((32 >> vl) * (vn + 1)) / 8. Terrain positions are vn=2, vl=1 (command 0x69) at 6 bytes. Reading the format from the immediate's low nibble -- the more commonly quoted layout -- yields 12 or 16 bytes and misaligns everything after it. Unpack payloads pad to 4 bytes, and several non-unpack commands carry inline data: STROW/STCOL 16, STMASK 4, MPG num*8, DIRECT/DIRECTHL imm*16. Ignoring those desynchronises the walk, and a desynchronised walk still finds code words that look like positions. Positions are also not all in one list. vert_count is the total across three tiers, the LOD-2 list precedes the common one in offset order, and the LOD-0-only list starts at shared_ofs + lod_1_size*16 rather than at an offset of its own. Scanning only the common list matches about a quarter of patches, each exactly -- which is what made it convincing. The invariant is that every patch yields exactly its own vert_count. It caught both errors, so tfrag_positions returns None on a mismatch rather than partial geometry. All 27 levels: 18,102 patches, 821,439 vertices, zero mismatches. Still open: strips, triangle indices and the RGBA vertex-info arrays. Positions alone are not drawable. |
||
|
ReaperOfVeriod
|
16c04db559 |
Fix level extents read as bytes, and extract the level geometry
A defect in the Phase 1 extractor, found by reading the data extent's contents: level extents are declared in sector counts, the generic G/ pair table uses byte counts, and level containers were being routed through the generic path. So LEVEL0.WAD's data extent -- 6,484 sectors -- was taken as 6,484 bytes. Every test still passed, because a truncated file is still a file. The extent's own header gave it away: it declares an uncompressed total of 13,277,920 bytes, and 13,277,920 / 6,484 is 2047.7. A ratio landing on 2048 is a unit error, not a compression ratio. Across 27 levels this had been discarding roughly 300 MB of level content. Level containers now go through the typed LevelHeader instead. Level 0 yields: data.bin 13,279,232 the data extent core_data.wad 9,738,678 the geometry, compressed core_data.bin 16,969,296 the geometry, decompressed core_index.bin 31,600 the geometry directory gameplay.bin 659,456 sound_bank.bin 1,503,232 core_data is a WAD LZ77 stream -- the format already implemented for the game -- and it decompresses cleanly. The first real level content this project has recovered. New: LevelDataHeader parses the data extent's own header, a table of byte ranges naming the level overlay executable, the geometry directory, the GS RAM image, the HUD banks and the geometry. Note this is a third meaning for a "size" field on the same disc, after sector counts in the level header and byte counts in the generic table; all three are documented together in level-data.md so the next parser does not rediscover this the hard way. The geometry directory's offsets were checked against the geometry rather than taken on trust: all twelve land inside the 16.9 MB of decompressed core_data, which is what confirms the header is being read correctly. The class tables themselves do not yet decode under the documented layout, and that is recorded as open work in plan/level-assets.md rather than papered over. 75 unit tests, 9 integration tests against the real disc. Hash table regenerated at 5,159 files and 5,671,915,799 bytes -- hashes only, no content. Extraction now writes to ~/Documents/gc-extracted rather than /tmp, which is 16 GB and did not have room for a 5.3 GB output. |
||
|
ReaperOfVeriod
|
a9e43e39c1 |
Parse the level container header, naming each level's four extents
The first typed structure read from real data. Every one of the 27 level containers begins with the same 0x60-byte header, which turns the generic pair table into named assets: data, sound_bank, gameplay and occlusion, plus three chunk descriptors. Two things about this header are worth stating because both are silent failures rather than errors. Sizes here are sector counts, not byte counts. The generic G/ pair table on the same disc uses byte counts, so both readings type-check and the wrong one is off by a factor of 2048. The tiling check settles it: read as sectors, the four extents tile every one of the 27 containers exactly; read as bytes they overlap each other. Level id is not container order. LEVEL21.WAD reports id 30, so a level was cut from the disc and its number left a gap. Anything keying off file order -- a save slot, the planet select screen -- would mislabel that planet. There is also a second trap in the same area: the ISO filesystem sorts names alphabetically, so LEVEL10 precedes LEVEL2, and any tool needing order has to sort numerically. The header's field at 0x04 is zero on all 27 levels. It is not the container size despite looking like one, so the code reads it as a runtime field and takes the container length from the ISO entry instead. Measured across all 27 levels, three invariants now hold in tests: extents tile in file order; extent coverage plus chunks reaches at least the end of the container (21 levels land exactly on its length, 6 leave 134 to 689 sectors of trailing data); and exactly one level's id differs from its container index. The coverage assertion is >= rather than == on purpose -- 21 levels end one sector past the file end because the header occupies sector 0, and asserting equality would truncate a level's tail. gameplay is left unparsed and that is recorded rather than hidden. It begins "WAD" plus a per-level byte and does not decode with the WAD LZ77 decoder: all 27 levels produce plausible output for the first packet and then fail on a rule the format's own encoder never breaks, which means the input is wrong rather than the decoder. docs/plan/level-assets.md records what is ruled out and what to try next. 66 unit tests, plus 7 integration tests against the real disc. The unpack hash table is unchanged, so this adds parsing without altering the pipeline. |
||
|
ReaperOfVeriod
|
027373e65b |
Add the disc pipeline: identify, unpack and verify
Phase 1 of the roadmap. The disc is now turned into verified asset files, and
the result is checked against a committed hash table rather than against
whatever happens to be on disk.
gc-extract identify disc.iso SCUS_972.68 v1.01 NTSC, ELF SHA-1 matches
gc-extract unpack disc.iso out/ 5135 files, 4,500,106,960 bytes, 10.8s
gc-extract verify out/ table.tsv OK: 5135 files match
All 92 G/*.WAD containers parse. MPEG1 and MPEG2 are raw video and pass
through whole; the other 90 are indexed containers whose extents are extracted
individually.
Formats recovered and written up in docs/formats/:
- iso9660.rs. R&C2 lists its assets in the filesystem, unlike R&C1, so a
directory walk is the primary asset index. Three details that all fail
quietly rather than loudly: directory records store every 32-bit field twice
(little-endian then big-endian), the CD001 signature is at sector offset 1
rather than 0x8001, and the directory flag at record offset 25 is
authoritative where the identifier's trailing slash is not.
- global_wad.rs. Container headers hold a table of extents. All 92 indexed
containers use {offset in sectors, size in bytes}. Measured, not assumed: the
sector-count reading places every entry of all 27 LEVEL<n> and all 27
AUDIO<n> containers out of range, while the byte reading places all of them
correctly. SCENE<n> is the one exception, a flat sector list whose extents
are sized by the next value.
- wad.rs. Insomniac's LZ77 stream: four flag-selected packet classes, with 0-3
trailing literal bytes per packet carried in unused low bits. This is not a
bit stream, which is why it resists the usual LZ77 reading.
- buf.rs. A bounds-checked reader, so untrusted 2003 disc data produces typed
errors rather than panics.
gc-extract carries the three stages and a SHA-1 implementation with FIPS 180-4
test vectors, so identification works without pulling in a crypto dependency.
Verification: 54 unit tests, plus four integration tests that run only when
GC_DISC names a disc. Those four check identification against the recorded
values, cross-check the Rust identifier against tools/identify-disc.py, parse
all 92 containers, and confirm a full unpack matches the committed table. The
committed table is 316 KB of sizes and SHA-1s for all 5,135 files: hashes only,
no game content, so it is safe to version and it turns "matches retail" into
something a test can assert.
docs/plan/disc-pipeline.md records two wrong turns, because both are easy to
repeat and both failed by producing plausible wrong output rather than an
error: the container table was read backwards, and the LZ77 stream was modelled
as a bit stream. The first survived because it happened to be correct on the
one container used for manual checking.
|