skate-3-rust-engine/docs/installation.md

Windows installation

scripts/Build-Release.ps1 produces target/skate3rust-windows-x64.zip. Keep its support directory beside skate3rust.exe. The game links Bevy and the MSVC runtime statically; the setup helper bundles Python, NumPy, Pillow and Tcl/Tk. It also bundles a small Rust RefPack decoder. Users need no compiler.

On first launch and after an extractor update, setup asks for a local Xbox 360 Skate 3 default.xex or ISO. Selecting default.xex uses the surrounding folder, which must still contain the full game data. Extracted copies skip ISO extraction and need no tool download. For ISO input, setup downloads the hash-pinned extract-xiso utility from XboxDev. Asset conversion runs in the bundled setup helper. Blender is neither required nor downloaded. No game files are downloaded or included in the package.

The conversion pipeline:

1. Extracts the disc into a new installation workspace.
2. Preserves native animation banks and state graphs from miscload.big, and the controller mapping from miscboot.big.
3. Reads the big-endian 64-bit AttribSys schema and collections from db.big. Settings are decoded from the disc, with numeric hashes retained for names absent from debug strings. The JSON export serves the fields consumed by this engine; it is not a general VLT editor or portable array serialization.
4. Builds the default modular skater from the CAC archive, including textures, morph settings, the native rig and its four board-relative IK targets. Writes GLB directly with retail skin weights, inverse-bind matrices and the bone-local basis used by the runtime. Gameplay animations are evaluated from ABIN by the Rust runtime, without offline animation baking.
5. Converts each worldDIST_*.big directly into a v14 .skate map. Native collision archives and grind splines are embedded; collision triangles are decoded from that archive by the game rather than duplicated in the file. University uses the existing starting location. Other districts select a broad upward collision face near the district origin.
6. Runs the game's asset and map loaders before publishing installation.json. This validates loading, not gameplay. Conversion intermediates are removed after their output succeeds. Failed workspaces retain logs for diagnosis.

Data lives in data beside this portable copy's executable. installation.json points to the completed installation. Conversion tools are cached separately, so retrying does not download them again. setup-error.log reports setup errors; each installation also has setup.log with converter output.

For an already extracted disc, developers can use tools/prepare_assets.py with --game-root, --output and --game-exe. The release helper can run the same entry point with --task tools/prepare_assets.py.

The Escape menu discovers installed .skate files. Loading a map starts a new game process and closes the current session so physics and render resources are recreated. A fresh launch defaults to University. Development checkouts with assets/private/game.json retain their existing local assets and default test world; --assets DIRECTORY also selects a prepared asset set explicitly.

Conversion checks

The direct converter was checked against an extracted Xbox 360 disc containing the three main districts, Skate School, and six separately stored parks: Black Box, Downtown Skate Park, Industrial Skate Park, Maloof, Mega Park and Start Park. All ten outputs passed the runtime's asset, spline and collision load checks. This does not establish visual or gameplay parity, and the original ISO extraction step still needs checking with an ISO.

Two retail spline details needed corrections. Downtown contains duplicate knots and sub-millimetre chords. The native cubic records remain intact; only unusable contact primitives are omitted. Skate School's first spline header word is 0001 0002, meaning two rails, not 65,538. TU3 function 82C1EEF0 reads the count with lhz +2 (also at 82C1EFBC). The reader now uses that halfword and retains the other halfword in the map metadata.

Setup performance

On the development PC, a full packaged conversion of the same extracted disc dropped from approximately 13 minutes to 81 seconds. This includes skater preparation, all ten maps, runtime load checks and cleanup; ISO extraction is excluded. Timing depends on CPU, available memory and storage.

Setup uses native RefPack decompression, batched NumPy texture decoding, uncompressed temporary geometry/texture caches and faster lossless final compression. The spawn search evaluates triangles in batches and rejects distant collision bounds before decoding them. Up to three isolated map workers run together, limited by available RAM, with the largest districts scheduled first. Per-map conversion and load logs remain in the installation.

Decoded map geometry, material data, texture pixels, native collision archives, rail payloads and the character GLB were compared with the earlier outputs. Final map storage uses a different compression level, without reducing quality.

Separate portable installations and asset refresh

Each copy uses data/installation.json beside its executable. A freshly unpacked ZIP has no record and always asks for an Xbox source. Startup does not search the working directory, another package, or %LOCALAPPDATA% for prepared game assets. Once that copy is set up, ordinary launches reuse its own data. --assets remains an explicit developer override and bypasses setup and asset-refresh management. Do not ship a local data folder in release ZIPs.

The release manifest includes fingerprints for core data, HUDs, character, environment and maps. They are derived from the packaged extractor sources and shared dependencies, not the game build number. An in-place program update keeps data intact. Next startup offers an asset refresh only when these fingerprints change. A HUD extractor edit refreshes HUDs without converting maps; changes to core VLT/animation decoding invalidate dependent groups too. Shared map/material parsers can invalidate both maps and environment assets. New converter modules must be assigned to their consuming groups in asset_pipeline/versions.py.

Refresh uses the same Select ISO or default.xex file picker as initial setup; the player never needs to select an intermediate extraction folder. ISO sources must be unpacked again when a refresh needs disc files; unchanged asset groups are reused without conversion. The chosen Xbox executable must match the original edition. Source paths stay in local installation records and are never included in published release metadata.

Updates run only affected exports. Core/HUD/character/environment/map recipes have separate dependency identities; setup UI and transaction changes do not invalidate every map. Explicit, exact old/new fingerprint equivalences cover verified unchanged exports when migrating the earlier broad fingerprint scheme. Unrecognized versions still refresh normally.

Core refreshes create a new installation before publication. Unchanged prepared maps and immutable character generations use hardlinks (copy fallback on filesystems without support); mutable user files and outputs being rebuilt have independent storage. Customiser-only refreshes keep the current core installation and publish a new character generation. Custom models, profiles, settings, selections and mods are retained. Every required stage, including the complete character library and pro roster, must succeed before the installation record changes. Failures preserve the previous record. A kernel lock prevents concurrent setup and releases automatically on process exit, including a crash; the harmless setup.lock file can remain on disk.

Character catalog, library, menu, lighting and roster stages have separate source/dependency identities and SHA-256 output receipts. Completed stages can be reused across generations or resumed after an interruption; partial stages are rebuilt before any existence-based converter sees them. Setup verifies checksums before reuse. Startup checks receipt paths and file sizes, avoiding a full map checksum scan on every launch. Same-size corruption is detected during setup validation, rather than ordinary startup. Old generations remain for recovery; this still requires space for genuinely changed output.

See character preparation measurements for benchmark conditions, stage timings and output equivalence checks.

Installations from releases that used the old global asset store need one setup in the new per-copy layout. There is deliberately no automatic global migration. Manually unpacking a ZIP over the same folder retains that folder's data and behaves like an in-place update; unpack into a new folder for a fresh setup.