5,389
edits
OBD:Localization/Japanese: Difference between revisions
m
no edit summary
(basic notes on the (a?) Japanese release; to be continued) |
mNo edit summary |
||
| Line 54: | Line 54: | ||
Apart from the localization, the Japanese level0_Tools offers better quality TXMP than the German Big Blue Box one, on 91 textures out of 373: DOORLOCK, DOORLOCK2, and all those with prefixes DOOR_AIR, DOOR_MP, DOOR_SB, _DOOR_AIR_, _DOOR_DR_, _DOOR_HA_, _DOOR_MP_, _DOOR_PW_, _DOOR_PW2_, _DOOR_RL_, _DOOR_SB_, _DOOR_TC_, _DOOR_TC2_, _DOOR_Z_, A_GARAGE_, AIR_DK_, CON_SB_ (uncompressed storage rather than DXT1). See HERE for an overview of level0_Tools that is not focused on localization. | Apart from the localization, the Japanese level0_Tools offers better quality TXMP than the German Big Blue Box one, on 91 textures out of 373: DOORLOCK, DOORLOCK2, and all those with prefixes DOOR_AIR, DOOR_MP, DOOR_SB, _DOOR_AIR_, _DOOR_DR_, _DOOR_HA_, _DOOR_MP_, _DOOR_PW_, _DOOR_PW2_, _DOOR_RL_, _DOOR_SB_, _DOOR_TC_, _DOOR_TC2_, _DOOR_Z_, A_GARAGE_, AIR_DK_, CON_SB_ (uncompressed storage rather than DXT1). See HERE for an overview of level0_Tools that is not focused on localization. | ||
==The size of the .dat files== | |||
It shouldn't be ''too'' surprising that 4 out of 15 .dat files (not counting level0_Tools) have exactly the same size as their English counterparts. If two level files contain exactly the same instances (possibly in a different order), and if matching instances differ in content only but not in size (e.g., with translated TxtC/OPge/IPge/etc, but no size change for array instances like TXCA/VCRA/IDXA), then the .dat files' sizes will naturally match. Textures, sounds and subtitles are likely to vary for a localization, but their variable data resides in the .raw, and does not affect the size of the .dat files. | |||
So the real question, rather, is why would the other eleven .dat files be larger than their English counterparts? | |||
===level0_Final=== | |||
The case of level0_Final is special: its much larger size (4,520,626 B against 2,816,050 B) is almost entirely attributable to the font files (TSFT) and glyph arrays (TSGA) belonging to the Tahoma TSFF. | |||
Like for the English Oni, there are fifteen TSFT font files (normal, bold and italic variants in 7pt, 9pt, 10pt, 12pt and 14pt sizes). Each of them has a slightly larger pixel data array: there are no extended ASCII glyphs (96 of those), but the 154 kana more than make up for them, so the pixel data of the TSFTs ends up taking more than for English Oni, by 230,560 B overall. | |||
The main difference is size comes from the somewhat bloated use of auxiliary TSGAs for kana descriptors: in addition to the main TSGA used for regular ASCII (less that half full), there are 19 more TSGAs, totalling space for 4864 glyph descriptors (per TSFT), but filled very sparsely (only 154 kana glyphs are actually present). The excess space taken up by the 19 auxiliary TSGAs is about 100 kB (including instance headers, instance descriptors, and padding), so for the 15 fonts we get about 1.5 MB (about 30 times more than for the glyphs themselves, i.e., the pixel data in the TSFTs). | |||
===Actual levels=== | |||
Typically a level will have larger .dat file for the Japanese version because it contains more instances, and these extra instances are typically orphan ones (not actually looked up by the game, or the original Oni would require them too), left over from the development/localization process. Remember that even the original Oni has orphan instances, both named (unreferenced SNDDs, such as Iron Demon sounds) and unnamed (like the low-LOD TRCM that are bundled with animation collections, see [[OBD:File_types/Named#TRCM|HERE]]). | |||
A detailed overview of unnamed orphan instances is somewhat laborious, since those are lost upon -export (unpacking via OniSplit), and can only be analyzed through OniBrowser and/or in binary. However, their abundance (relative to the English version) can be estimated by repacking each level and seeing by how much it shrinks. | |||
Named orphans can be assessed directly. In the case of the Japanese version, the extra named instances are limited to alternative takes of speech lines (SNDD files). | |||
==level#_Final textures== | ==level#_Final textures== | ||
| Line 75: | Line 94: | ||
The following WMM_s and WMDDs were ''not'' translated: | The following WMM_s and WMDDs were ''not'' translated: | ||
==AKEVtctf== | |||
Somewhat surprisingly, there is a minor change in AKEVtctf, namely UV coordinate #14836 in TXCA: | |||
:in English Oni the value is (7.35236319953e-30, 7.08227594387e-39), or (0x0F151FA7,0x004D1E81) in hex | |||
:in Japanese Oni the value is (8.97938881655e-30, 7.08301582946e-39), or (0x0F'''36'''1FA7,0x004D2091) in hex | |||
For all intents and purposes both UVs are equivalent to (0,0). The reason for the diff is anyone's guess. | |||
==Subtitles== | ==Subtitles== | ||
SUBTsubtitles and SUBTmessages were fully translated, with the exception of two unused voice lines: | SUBTsubtitles and SUBTmessages were fully translated, with the exception of two unused voice lines: | ||
| Line 104: | Line 126: | ||
*'''city_amb2_t07b''' lasts ~5 seconds (319 ticks) in English Oni and ~9 seconds (536 ticks) in Japanese Oni | *'''city_amb2_t07b''' lasts ~5 seconds (319 ticks) in English Oni and ~9 seconds (536 ticks) in Japanese Oni | ||
*'''city_amb2_t07c''' lasts ~4 seconds (215 ticks) in English Oni and ~14 seconds (853 ticks) in Japanese Oni | *'''city_amb2_t07c''' lasts ~4 seconds (215 ticks) in English Oni and ~14 seconds (853 ticks) in Japanese Oni | ||
Thus, of all the "city ambient" loops, only ''' | Thus, of all the "city ambient" loops, only '''city_amb_t06a''' uses the same sound data as in English Oni. | ||
---- | ---- | ||
In EnvWarehouse (level 1) there are backups of three English Oni sounds, not referenced in OSBD: | In EnvWarehouse (level 1) there are backups of three English Oni sounds, not referenced in OSBD: | ||
| Line 112: | Line 134: | ||
---- | ---- | ||
Some recurrent sounds are different across levels in Japanese Oni, instead of being exact duplicates. | Some recurrent sounds are different across levels in Japanese Oni, instead of being exact duplicates. | ||
*'''c00_01_29shinatama''' ("Hi Konoko! It's me, Shinatama." | *'''c00_01_29shinatama''' occurs both in level 0 and level 1, as two takes with slightly different intonations (the English counterpart, "Hi Konoko! It's me, Shinatama.", is exactly the same in level 0 and level 1). | ||
---- | ---- | ||
| Line 126: | Line 148: | ||
*snin_hrt_md4 (different from the actually referenced "snin_ hrt_md4" as well as from its original English version) | *snin_hrt_md4 (different from the actually referenced "snin_ hrt_md4" as well as from its original English version) | ||
*snin_hrt_md5 (different from the actually referenced "snin_ hrt_md5" as well as from its original English version) | *snin_hrt_md5 (different from the actually referenced "snin_ hrt_md5" as well as from its original English version) | ||
*c09_31_25shinatama (different from the actually referenced "c09_31_25shinatama " as well as from its original English version) | *c09_31_25shinatama (different from the actually referenced "c09_31_25shinatama " as well as from its original English version; the reference is from OSBDc09_31_25shinatama_.grp, with an underscore character instead of the space) | ||
To be continued... | To be continued... | ||