Extracting NES and N64 ROMs from The Legend of Zelda: Collector's Edition (GameCube)

[RealRelativeEase]

I stumbled upon a blogpost by @sethmlarson which makes the extraction process for the Zelda Collection which had been released on GameCube a lot easier. Just download this script, put it in the same folder as the .tgc files extracted via the Dolphin file system browser and run python [name of script].py. The script will extract all of the ROMs, No-Intro names and iNES 2.0 header included.

I took the liberty of making an edit to the script to make it compatible with the PAL version of the compilation, updating the No-Intro names in the process:

# License: MIT

import hashlib
import os.path


def extract_nes(data: bytes, length: int) -> bytes:
    offset = data.find(b"NES\x1A")
    if length == 0x20000:
        header = b"\x4E\x45\x53\x1A\x08\x00\x12\x08\x00\x00\x70\x07\x00\x00\x00\x01"
    elif length == 0x40000:
        header = b"\x4E\x45\x53\x1A\x08\x10\x12\x08\x00\x00\x70\x00\x00\x00\x00\x01"
    else:
        raise ValueError(f"Unknown header for ROM length: {length}")
    offset += len(header)
    return (
        header +
        data[offset:offset+length]
    )


def extract_n64(data: bytes, length: int) -> bytes:
    offset = data.find(b"\x80\x37\x12\x40")
    return data[offset:offset+length]


def main():
    for to_file, from_tgc, rom_type, length in (
        # ROM names and lengths from No-Intro: https://datomatic.no-intro.org/
        ("Legend of Zelda, The (USA) (Zelda Collection).nes", "EUR_NES_ZELDA1.tgc", "NES", 0x20000),
        ("Zelda II - The Adventure of Link (USA) (Zelda Collection).nes", "EUR_NES_ZELDA2.tgc", "NES", 0x40000),
        ("Legend of Zelda, The - Ocarina of Time (Europe) (GameCube).z64", "zelda_PAL_093003.tgc", "N64", 0x2000000),
        ("Legend of Zelda, The - Majora's Mask (Europe) (GameCube).z64", "majora_PAL_100403c.tgc", "N64", 0x2000000),
        ("Legend of Zelda, The (USA) (Zelda Collection).nes", "USA_NES_ZELDA1.tgc", "NES", 0x20000),
        ("Zelda II - The Adventure of Link (USA) (Zelda Collection).nes", "USA_NES_ZELDA2.tgc", "NES", 0x40000),
        ("Legend of Zelda, The - Ocarina of Time (USA) (GameCube).z64", "zelda_ENG_090903.tgc", "N64", 0x2000000),
        ("Legend of Zelda, The - Majora's Mask (USA) (GameCube).z64", "majora_ENG_091003.tgc", "N64", 0x2000000),
    ):
        if not os.path.isfile(from_tgc):
            print(f"File not found: {from_tgc}")
            continue
        with open(from_tgc, "rb") as f:
            data = f.read()
        if rom_type == "NES":
            rom = extract_nes(data, length)
        elif rom_type == "N64":
            rom = extract_n64(data, length)
        else:
            raise ValueError(f"Unknown ROM type: {rom_type}")
        rom_md5 = hashlib.md5(rom).hexdigest()
        with open(to_file, "wb") as f:
            f.truncate()
            f.write(rom)
        print(f"Extracted {to_file} ({rom_md5})...")


if __name__ == "__main__":
    main()

I don't have a Japanese copy of the collection, so I can't provide an edit for that release. But I would assume that it uses the Famicom Disk System versions instead of the Western releases.

[RealRelativeEase]

I also tried to make another edit to the Python script to extract the N64 ROMs from the PAL version of the Wind Waker bonus disc:

# License: MIT

import hashlib
import os.path


def extract_all_n64(data: bytes, length: int):
    """Return all ROMs starting with the N64 magic header."""
    magic = b"\x80\x37\x12\x40"
    offsets = []
    start = 0
    while True:
        off = data.find(magic, start)
        if off == -1:
            break
        offsets.append(off)
        start = off + 4
    return [data[o:o+length] for o in offsets]


def main():
    # Expected MD5s
    expected = {
        "Legend of Zelda, The - Ocarina of Time (Europe) (GameCube).z64":
            "2c27b4e000e85fd78dbca551f1b1c965",
        "Legend of Zelda, The - Ocarina of Time - Master Quest (Europe) (En,Fr,De) (GameCube).z64":
            "1618403427e4344a57833043db5ce3c3",
    }

    for to_file, from_tgc, rom_type, length in (
        ("Legend of Zelda, The - Ocarina of Time (Europe) (GameCube).z64", "zlj_f.tgc", "N64", 0x2000000),
        ("Legend of Zelda, The - Ocarina of Time - Master Quest (Europe) (En,Fr,De) (GameCube).z64", "zlj_f.tgc", "N64", 0x2000000),
    ):
        if not os.path.isfile(from_tgc):
            print(f"File not found: {from_tgc}")
            continue

        with open(from_tgc, "rb") as f:
            data = f.read()

        if rom_type != "N64":
            raise ValueError(f"Unknown ROM type: {rom_type}")

        # Extract all possible ROMs
        candidates = extract_all_n64(data, length)

        # Find the one with the correct MD5
        target_md5 = expected[to_file]
        rom = None
        for c in candidates:
            if hashlib.md5(c).hexdigest() == target_md5:
                rom = c
                break

        if rom is None:
            print(f"Correct ROM not found for {to_file}")
            continue

        with open(to_file, "wb") as f:
            f.truncate()
            f.write(rom)

        print(f"Extracted {to_file} ({target_md5})...")


if __name__ == "__main__":
    main()

[aceofstax]

I don't have a Japanese copy of the collection, so I can't provide an edit for that release. But I would assume that it uses the Famicom Disk System versions instead of the Western releases.

I do, and I can test it sometime soon. I know I have the container files from the disc backed up, but I don't remember if I had extracted the Famicom ROMs yet. I'll post back later.

EDIT: correcting a statement previously made about the N64 ROMs and the .tgc files... They're containers with .n64 files inside them. They can be accessed by putting the .tgc files in your Dolphin emu games folder, right click and go to properties, then go to the Filesystem tab.

At least on the US version I recall using some .NES extraction python script well known to RED-Project. I seem to remember extracting a Famicom rom with it too, but I had to use a fork or modify the script.

FWIW, I have the JP Master Quest disc too, and the N64 ROMs are stored the same way as the Zelda CE disc.

EDIT: JanErikGunnar's vcromclaim nes_extract.py seems to have the logic for identifying FDS ROMs. "NES" is not found in these ROMs, but rather this instead:

*NINTENDO-HVC*

[Shinyun226]

Not sure how much this will help but... I have the Japanese version and extracted the Disk System roms years back -
JPN_JPN_ZELDA0.tgc and JPN_JPN_ZELDA2.tgc are Zelda 1 and 2 respectively, but you have to mess with them a bunch before they'll be usable in an emulator.
Essentially I had to:

1. Extract those tgc files from the disk image

2. Extract default.dol from those tgce files using tgc extractor

3. Use dolsplit to get all of the data#.bin files out of default.dol

4. Extract the qd files (zelda_6.qd and link_3.qd) out of data4.bin using arcextract
   (edit: nes_extract.py, mentioned above, probably takes care of steps 1-4 if it's finding the "NINTENDO-HVC" header)

5. Then finally use qd2fds.py to convert the .qd files to .fds
   (I believe this is the site I used to find most of those steps back then for reference: https://ayasuke.exblog.jp/7983954)

The resulting Zelda 1 FDS rom works perfectly, however the Zelda 2 rom will freeze anytime you die - I'm guessing this is because the flashing that originally occurred here was removed. The built-in emulator on the collection disk must compensate for this somehow, but the various nes emulators I tested + the MiSTer FPGA NES core would all freeze here.
I'm pretty sure I ended up using a hex editor to fix that somehow, but honestly don't remember how I accomplished all that.

edit: Turns out I still have both versions of the file. Compared them in a hexeditor and looks I made a minor edit at offset 1DC3.
Changed "4C 13 EB" to "29 03 AA"; that restores the flashing effect and allows the game to continue after you die.

[RealRelativeEase]

I do, and I can test it sometime soon. I know I have the container files from the disc backed up, but I don't remember if I had extracted the Famicom ROMs yet. I'll post back later.

I'm sure you probably already know, but the N64 ROMs just need a simple filename and filetype rename. At least on the US version I recall using some .NES extraction python script well known to RED-Project. I seem to remember extracting a Famicom rom with it too, but I had to use a fork or modify the script.

FWIW, I have the JP Master Quest disc too, and the N64 ROMs are stored the same way as the Zelda CE disc.

That would be great, I guess there's no point in trying to separate these scripts by region. If you can give me the file names for the ROMs and .tgc files, it should be trivial to add at least the N64 ROMs.

Not sure how much this will help but... I have the Japanese version and extracted the Disk System roms years back - JPN_JPN_ZELDA0.tgc and JPN_JPN_ZELDA2.tgc are Zelda 1 and 2 respectively, but you have to mess with them a bunch before they'll be usable in an emulator. Essentially I had to:

1. Extract those tgc files from the disk image

2. Extract default.dol from those tgce files using [tgc extractor](http://nekokabu.s7.xrea.com/soft/dcl.cgi?act=up&url=tgce_06.lzh)

3. Use [dolsplit](http://nekokabu.s7.xrea.com/soft/dcl.cgi?act=up&url=dolsplit_01.lzh) to get all of the data#.bin files out of default.dol

4. Extract the qd files (zelda_6.qd and link_3.qd) out of data4.bin using [arcextract](http://nekokabu.s7.xrea.com/soft/dcl.cgi?act=up&url=arce_02.lzh)
   (edit: nes_extract.py, mentioned above, probably takes care of steps 1-4 if it's finding the "_NINTENDO-HVC_" header)

5. Then finally use [qd2fds.py](https://gist.github.com/infval/18d65dd034290fb908f589dcc10c6d25) to convert the .qd files to .fds
   (I believe this is the site I used to find most of those steps back then for reference: https://ayasuke.exblog.jp/7983954)

The resulting Zelda 1 FDS rom works perfectly, however the Zelda 2 rom will freeze anytime you die - I'm guessing this is because the flashing that originally occurred here was removed. The built-in emulator on the collection disk must compensate for this somehow, but the various nes emulators I tested + the MiSTer FPGA NES core would all freeze here. I'm pretty sure I ended up using a hex editor to fix that somehow, but honestly don't remember how I accomplished all that.

edit: Turns out I still have both versions of the file. Compared them in a hexeditor and looks I made a minor edit at offset 1DC3. Changed "4C 13 EB" to "29 03 AA"; that restores the flashing effect and allows the game to continue after you die.

I'm no programmer, so I can't make any promises. But I'll see if I can give this a try when I have the time.

[aceofstax]

That would be great, I guess there's no point in trying to separate these scripts by region. If you can give me the file names for the ROMs and .tgc files, it should be trivial to add at least the N64 ROMs.

Sure, it's no problem. I do have to correct a previous statement I made-- the .tgc files aren't just a simple rename. I may have confused this with an experience archiving N64 ROMs from a Wii title.

The ROMs on the Japanese disc are split between two folders. I don't recall this being the case on the US version, so it seems a little odd.

Anyway here are the two significant folders and their file listings:

PZLJ01 (Zelda Collection JP)

• tgc
  • 120903_zelda.tgc (contains zlj_f.n64)
  • JPN_JPN_ZELDA0.tgc
  • JPN_JPN_ZELDA2.tgc
  • retrospec_movie_JP_final02.tgc
  • windwaker_movie_JP_final02.tgc
  • ZL_WindWakerSHOP_final_2003-10-31_17-54-53.tgc
• tgcmj
  • 120903_majora.tgc

D43J01 (Zelda Master Quest JP)

• zlj_f.tcg (contains zlj_f.n64 and urazlj_f.n64)

From here you can extract the ROMs via GUI through Dolphin. Dolphin is how I got the .tgc files, you can browse the .tgc file contents by dropping it in the same folder as your GameCube/Wii .iso files. This works for only the N64 ROMs however-- So manually extracting the FDS roms is necessary for the Famicom games.

Due to read errors on my Zelda Collection disc, I can only provide info on all .tgc files mentioned with exception of Majora's Mask (I need to repair the disc before I can get a proper read...)

Zelda Collection JP N64 ROMs:

• 120903_zelda.tgc
  • zlj_f.n64

Zelda no Densetsu - Toki no Ocarina (Japan) (GameCube, Zelda Collection).z64

• 120903_majora.tgc
  • [original filename to be added later]

Zelda no Densetsu - Mujura no Kamen (Japan) (GameCube).z64

Master Quest JP N64 ROMs:

• zlj_f.tgc
  • zlj_f.n64

Zelda no Densetsu - Toki no Ocarina GC (Japan) (GameCube).z64

• urazlj_f.n64

Zelda no Densetsu - Toki no Ocarina GC Ura (Japan) (GameCube).z64

So the hashes between the two different Japanese OOT roms aren't the same (Zelda Collection vs Master Quest). Note that only Master Quest JP disc uses "GC" in the No-Intro filenames. IIRC, the GC appears on the in game title screen.

[aceofstax]

EDIT (2026 Apr 09): Post has been updated & corrected.

I have an update on my own effort in extracting these ROMs. My approach was a little different but it seems I may have achieved the same end result.

However, I did not end up using nes_extract.py. Actually it's possible to extract the main.dol of both tgc archives via Dolphin.

I'll go into detail further down, but the resulting files match known No-Intro entries without further modification. Unsure about playability. I did run into an error, but it could have been bad emulator configuration on my end.

Not sure how much this will help but... I have the Japanese version and extracted the Disk System roms years back -
JPN_JPN_ZELDA0.tgc and JPN_JPN_ZELDA2.tgc are Zelda 1 and 2 respectively, but you have to mess with them a bunch before they'll be usable in an emulator.
Essentially I had to:

1. Extract those tgc files from the disk image
2. Extract default.dol from those tgce files using tgc extractor
3. Use dolsplit to get all of the data#.bin files out of default.dol
4. Extract the qd files (zelda_6.qd and link_3.qd) out of data4.bin using arcextract
   (edit: nes_extract.py, mentioned above, probably takes care of steps 1-4 if it's finding the "NINTENDO-HVC" header)

I actually did not have luck using the nes_extract.py script-- it seems that it simply found the .qd files and added an FDS header to them. I'm not entirely sure as I'm not a coder. The folks at No-Intro allegedly extracted these with szstools, but I didn't have any luck with that.

Here's what I ended up doing instead:

To extract these .tgc files via Dolphin GUI, you have to add them to the games folder registered in Dolphin settings (basically, put the .tgc files in with your GC .iso files).

Right click the titles (1.63MB per .tgc file), go to Properties (.tgc filename should appear in the new window's title bar), click the Filesystem tab, then right click the top of the filesystem tree where it says "Disc - PZLJ01" and select "Extract System Data...". Save it to a folder named after the tgc file you're extracting from so you don't get them mixed up.

Extracting the system data will provide files not shown by Dolphin's GUI, saved to a subfolder named "sys". Inside you'll find a main.dol, and the .qd files can be found at the following specific offsets via HxD:

JPN_JPN_ZELDA0.tgc's main.dol:

Zelda no Densetsu - The Hyrule Fantasy (Japan) (Collector's Edition).qd
Found @ Block(h): 5E9A0-7E99F
CRC32: b9367eff

JPN_JPN_ZELDA2.tgc's main.dol:

Link no Bouken - The Legend of Zelda 2 (Japan) (Rev 1) (Collector's Edition).qd
Found @ Block(h): 5EA00-7E9FF
CRC32: df122ed1

Use qd2fds.py and you'll have the .fds versions:

Zelda no Densetsu - The Hyrule Fantasy (Japan) (Collector's Edition).fds
CRC32: b64b1456
Link no Bouken - The Legend of Zelda 2 (Japan) (Rev 1) (Collector's Edition).fds
CRC32: d9d60599