Pack format (inside TIM_LIST and TMD chunks) Confirmed
The simplest way Legaia bundles several assets into one blob: a count, a table of offsets, then the assets packed back-to-back. This shape appears as the data payload of a TIM_LIST or TMD chunk inside a DATA_FIELD streaming container (the game's typed-chunk multi-asset file format). Beware: it is one of three similar-looking pack layouts on the disc, each with different header math - the distinction section below tells them apart.
Overview
Implementation: crates/asset/src/pack.rs.
Layout
u32 count
u32 word_offset[count] // each is in 4-byte words from the start of THIS pack data
... sub-asset bytes, packed back-to-back ...Sub-asset i lives at byte range [word_offset[i] * 4 .. word_offset[i+1] * 4), with the last sub-asset ending at the chunk's end. Note the offsets count 4-byte words, not bytes.
Example
A TIM_LIST chunk header followed by a 2-TIM pack (a TIM is the PlayStation's standard texture format):
chunk header: 6c 02 01 01 type=0x01 (TIM_LIST), size=0x01026C
pack count: 02 00 00 00 count = 2
offset[0]: 03 00 00 00 word offset 3 → byte 12 (= start of TIM 0)
offset[1]: 8b 20 00 00 word offset 0x208B → byte 0x822C (= start of TIM 1)
[then 2 PSX TIMs back-to-back]Distinction from the standalone TIM-pack
The standalone TIM-pack form - used for whole archive entries rather than chunks inside a stream - is structurally similar but uses an 8-byte header (a byte[3] == 0x01 / byte[2] < 0x10 discriminator pair, then a u32 count at +4) and adds a constant +4 to each word offset. The pack-inside-streaming-chunks form here lacks the discriminator prefix and its word offsets are taken straight from the pack-data start - use the right format for the source. (The third sibling, field-pack, is a magic-prefixed scene bundle with its own page.)