1
0
Fork 0
mirror of https://github.com/notwa/mm synced 2024-11-04 22:49:03 -08:00
majora's mask scripts, and some ocarina stuff too
Find a file
2015-05-26 08:43:23 -07:00
Lua abstract brush again, update titles 2015-05-26 08:43:23 -07:00
.gitattributes NEVER AGAIN. 2015-03-08 09:52:19 -07:00
.gitignore dump ordered tables to be git-friendly 2015-05-01 11:29:43 -07:00
chksum.py handle owl save sizes properly 2015-03-02 07:12:11 -08:00
heuristics.py remove actor heuristics for now 2015-03-13 12:16:26 -07:00
mm-bitflags.txt readme 2015-02-19 11:25:44 -08:00
n64.py rewrite functions in pyrex for performance 2015-03-04 07:21:18 -08:00
n64_fast.pyx rewrite functions in pyrex for performance 2015-03-04 07:21:18 -08:00
README.md Update README.md 2015-05-03 07:39:13 -07:00
util.py use debug ROM filenames, less verbose 2015-03-03 00:40:28 -08:00
Yaz0.py rewrite functions in pyrex for performance 2015-03-04 07:21:18 -08:00
Yaz0_fast.pyx rewrite functions in pyrex for performance 2015-03-04 07:21:18 -08:00
z64dump.py rewrite functions in pyrex for performance 2015-03-04 07:21:18 -08:00

Miscellaneous Zelda 64 Resources

i like to muck around in the memory of these games.

Majora's Mask

for brevity, all addresses written here are given for the original US version. refer to the spreadsheets or Lua tables for their equivalent in other versions.

spreadsheets

some sheets have been put together to dump data in. some of them can be used to predict the result of glitches.

  • Event Flags are being documented here.

  • Memory Addresses updates more frequently than MM addrs.lua.

  • Get Item Manipulation
    mzxrules did the original OoT one, i just jammed in MM's data for the item table and chest contents. spoilers: no desirable results besides light arrows, if it were even possible. potential crashes are not taken account for.

  • Entrance Data
    this is a huge laggy mess that brings google docs to its knees. enter an "exit value" in hex and it'll figure out exactly where it takes you. the three known wrong warps are in bold.

save files

save files are just memory dumps of some important data. save data is loaded into 801EF670, and extends roughly 0x4000 bytes.

in the versions with owl saves, regular saves are 0x100C in size, and owl saves are 0x3CA0. owls use the extra space primarily to store the pictograph picture.

note that some values don't are reset when reading/writing save files, even owl saves.

the game checks a checksum, and for the text "ZELDA3". each slot has one backup copy of itself, though they don't seem to be used? if a slot is corrupted, it will show up as empty in the menu.

the checksum is a 16-bit sum of all bytes up to that point, allowing overflows. i've written a checksum-fixing program in python for bizhawk savefiles.

owl saves always have 0x24 added to their checksum for some reason.

save files are ordered as such, offset from the first:

  • 0x00000 - slot 1
  • 0x02000 - slot 1 backup
  • 0x04000 - slot 2
  • 0x06000 - slot 2 backup
  • 0x08000 - slot 1 (owl)
  • 0x0C000 - slot 1 backup (owl)
  • 0x10000 - slot 2 (owl)
  • 0x14000 - slot 2 backup (owl)

bizhawk save files, at the time of writing, have the first file offset to 0x20800. also, their byte order is wrong.

here's my usual process for hacking on save files:

alias revend='objcopy -I binary -O binary --reverse-bytes=4'
s="Legend of Zelda, The - Majora's Mask (USA).SaveRAM"
x=mm-save.xxd
revend "$s"
xxd "$s" > $x
$EDITOR $x
xxd -r $x > "$s"
./chksum.py $s
revend "$s"

save files (for download)

because no one likes first cycle.

sometime i'll bother checking what the bombers/lottery codes are for these.

bitfields

scene flags

two regions of 0x960 bytes are allocated for all the scene flags in the game. the first (801EFAE0) is loaded from save files, the second (801F35D8) is used for in-game changes. basically, edit the first for save hacking, and the second for in-game hacking.

each scene in the game uses 0x14 bytes of scene flags. this implies there's 0x78 possible scenes: 0x78*0x14 = 0x960.

the current scene's flags are always copied into the same place in memory. they appear in a different order than in save files, however.

(four bytes each)
803E8978 corresponds to offset 0x4 in the save file.
803E897C corresponds to offset 0x8.
803E8988 corresponds to offset 0x0.
803E898C corresponds to offset 0xC.
803E8994 corresponds to offset 0x10.

here's an incomplete document on some of link's bitfields for JP 1.0.