Skip to content

Written by @tellowkrinkle.

These scripts can be used to easily update the sharedassets0.assets file in Higurashi games. This can save a lot of clicking in UABE and string edits were not even possible without these scripts.

Note: All scripts require Python 3, you may need to replace calls to pip and python with pip3 and python3 if that's what they are on your computer.

There is also another tool written in Rust which can simplify the process even more and you might not even need to read all the details below.


Before using these scripts, run the following command to install the dependencies. If you run into an issue with decrunch then read bellow.

pip install numpy Pillow unitypack

UnityPack has dependency on something called decrunch which requires Visual C++ Build Tools 14.0 (on Windows) but can still fail to install even if you have them. We're not using the part that needs decrunch (it's only used to decompress certain formats of image assets) anyway so if you run into these issues, simply download this fork of unitypack where the dependency is removed and install if with python install from within its folder.


This script will allow you to edit text files embedded in the asset files of games. This includes most UI elements. requires three arguments, the .assets file that contains the text you want to edit, a JSON file listing the edits you want to make, and an output directory where the edits will be outputted. The JSON file should be formatted as an array of objects with the fields CurrentEnglish, CurrentJapanese, NewEnglish, and NewJapanese. There's also an optional Discriminator field, which shouldn't be used unless the script finds multiple copies of your string in the file, in which case it should tell you what values can be put there. Here's an example JSON file:

        "CurrentEnglish": "Original",
        "CurrentJapanese": "オリジナル",
        "NewEnglish": "PS3",
        "NewJapanese": "PS3"
    }, {
        "CurrentEnglish": "Updated",
        "CurrentJapanese": "新規",
        "NewEnglish": "MangaGamer",
        "NewJapanese": "MangaGamer"
    }, {
        "CurrentEnglish": "Watanagashi 12-3",
        "CurrentJapanese": "綿流し編 12日目3",
        "NewEnglish": "Watanagashi Day 12-3",
        "NewJapanese": "綿流し編 12日目3"

Run the program with python sharedassets0.assets changes.json outputFolder The program will then output one ###.dat file per change into the specified output folder, where the ### is the object ID of the file that it should replace in the assets bundle. These can be used directly with or manually added in using UABE.


This script will allow you to collect large numbers of asset bundle modifications into one large file that can be applied in one step using UABE.

EMIPGenerator requires three arguments, the .assets file that you want to edit, a directory of edits to make, and the name of the output file (should have the extension .emip or UABE won't let you load it).

The edit directory will be read as follows: Files that start with a number will replace the object with that number in the assets folder. For example, 2345.dat will replace object number 2345. 2345 Some helpful note as to what this file is.dat will also replace object number 2345. Files outputted by UnityTextModifier will be named properly to get picked up by this mode. Files that don't start with a number but are in the format name_type.dat will replace the first item in the bundle that has that name and type (for example, ConfigHigurashi_Texture2D.dat will replace the first item named ConfigHigurashi that is a Texture2D. Note that due to issues with UnityPack misreading files, the application uses a very messy way of searching files and therefore may not properly work with all file types in this mode. The first time you try to add a file in this mode, make sure the program doesn't print Couldn't find object named <name> for <filename>, skipping, which indicates that the script failed to find an object to replace. If this happens, manually look up the ID and prepend it to the file. Files that end in .png will be converted to Texture2D assets before they are inserted into the EMIP. Any file that does not start with a number or have an underscore in its name will be completely ignored * Any file that does have an underscore in its name but doesn't match a file in the asset bundle will still be ignored but may cause the program to run slower (due to searching the whole bundle for a matching file)

Run the program with python sharedassets0.assets inputFolder output.emip. Then, launch UABE and select File → Load Package File and select your emip file. If you get the error Unable to read the bundle file! (Invalid file or unknown version?), this means that you accidentally selected File → Open instead of File → Load Package File. Check the box next to the asset file on the following screen and if it doesn't show the full path to the file (starting with C: or whatever drive letter it's in), click the button near the bottom to select a base folder that will get the correct path. Then click OK, which will load the main bundle editing interface with all the edits already in. From here you can immediately select File → Save and save the output asset bundle.

UABE Screenshots

Select File → Load Package File and select your .emip file Load Package File Check the box and add a base folder if needed, then press OK Select a base folder Save the new bundle Save the new bundle


This script converts assets outputted by the current version of TextMeshPro into ones that can be read by the version used by the games. Note that if TextMeshPro updates it may stop working.

To use this, you will first need to generate a text atlas using TextMeshPro. Newer versions of Unity will output assets in a different format, so you will need an older one. 5.5.5 is known to work, while 2018.2 is known to not work. Due the python library having issues reading the assets outputted by TextMeshPro, you will need to then use UABE to extract the Atlas (Texture2D) and MonoBehaviour files from the asset, using the Export Raw button. If your version of Unity was too new, you'll find out here as UABE will fail to open the asset file.

You'll also need the MonoBehaviour file that you're trying to replace extracted from the game's asset bundle. The MonoBehaviour files contain pointers to other files in the assets so you can't reuse a file extracted from one chapter. For example if you want to replace the msgothic_0 font, use Export Raw on the msgothic_0 SDF file and save it as oldMonoBehaviour.dat:


Then, run python newAtlas.dat newMonoBehaviour.dat oldMonoBehaviour.dat outputFolder and the converter will create two files in the output folder with the same name as the two input files. These can be substituted back into the asset bundle. The atlas file can be left out if you don't need to convert it (if you already have a converted copy)

Getting character lists for font files

To make sure that the character sets you use have all the characters used by the games, use scripts in the CharacterInfoExtraction subfolder.

More detailed info on each of the scripts will be below, but assuming you compiled all the swift programs there and put them in the working directory, you can get a list of all the characters used in the Japanese script like this: cat pathToHigurashiDataFolder/StreamingAssets/Update/*.txt | ./HigurashiTextExtractor - j | ./UniqueCharacters - > japaneseScriptCharsUsed.txt. Note that you can get all the games at once by using multiple wildcards, for example if you have all the Mac versions in a folder you can do all of them at once with folder/Higurashi*.app/Contents/Resources/Data/StreamingAssets/Update/*.txt. Replace the j with e to get all the characters used in the English script. That's all you really need for the Japanese script (I recommend taking the kanji from that file and pasting them over the msgothic_0_charset_Japanese.txt in the scripts repo to make your final charset)

For English, you'll need to collect a few more files, since the font behind msgothic_2 will be used to display the game's menus, including the Japanese ones. Collect all the text used in strings in asset files using ./AssetBundleStringExtractor pathToHigurashiDataFolder/*.assets | ./UniqueCharacters - > assetBundleCharsUsed.txt. Then, clone the DLL code repository and cat all the Assets.Scripts.UI.Tips/TipsData.cs files along with the console arcs tips JSON together and run that through ./UniqueCharacters. Finally, you can cat englishScriptCharsUsed.txt assetBundleCharsUsed.txt tipsCharsUsed.txt | ./UniqueCharacters - > englishCharset.txt. Like with the Japanese, copy the kanji out of that and paste it over the kanji in the scripts repo's msgothic_2_charset_OtherLang.txt. This is to preserve the extra characters which were added to make translation into other languages easier.

Adding Font Support for a New Language

The first thing you'll need is a copy of Unity 5 (I used 5.5.5 but I would expect newer versions of Unity 5 to work as well). Once you have that, create a new project, go to the Asset Store, and search for and download TextMeshPro.

Preparing the character list

To make the font file, you will need a list of all the characters you want in it. Annoyingly, some of the Japanese parts of the games still rely on the English font, so you'll need to have those characters in addition to the ones you use for your language's character support. Download the current list from msgothic_2_charset_OtherLang.txt in here, which is the list of characters in the current English font. If your language's characters are already in that file, you shouldn't need to be doing font file editing at all. Otherwise, modify the text file to replace all the accented roman characters before the ←↑→↓ with all the characters you want in your font. Don't remove the characters after that point, because they're required to properly display some things in Japanese mode. Once you've done that, drag the new txt file into the Unity project you made.

Preparing the font file

Since you'll need support for the Japanese characters in addition to your own language, you'll need to add them to your font file. If your font file already has Japanese support (which may be the case for some Chinese or Korean fonts), you can skip this step. Otherwise, download FontForge and one of the TTF fonts here. Open both the target font and the one you just downloaded using FontForge. Navigate to Element → Font Info with both fonts, go to the General tab, and adjust the Em Size of one of the fonts to match the other (if you don't do this, one of the fonts will be way too small or big). Then, select Element → Merge Fonts... in the window for the target font to copy all missing glyphs from the Japanese font over. In addition, some fonts (like our current English font) use really thin lines for the ☆ character, which Higurashi uses a lot. Thin lines don't do very well with the font rendering system the game uses, which makes the character look really bad. Navigate to the entry for ☆, which you can do easily by selecting View → Goto and typing uni2606, and check if your font uses really thin lines. If it does, go to the Japanese font, go to the same spot, and copy and paste its ★ and ☆ over yours. Finally, export the new font to a ttf with File → Generate Fonts... and drag the font file into Unity.

Generating the SDF font

In your Unity project, choose Window → TextMeshPro → Font Asset Creator. Select your font as the font source, which should be there if you dragged it into your Unity project. For the font size, choose Custom Size. The current fonts use 26pt, but you can choose whatever you like. Bigger font sizes use more space but look nicer. Change Font Padding to 6 pt. If your characters almost fit in the atlas but not quite, you can change the packing method from Fast to Optimal to pack a few more characters in.

For the Atlas Resolution, larger numbers take up more space but allow more or larger characters. The current Japanese font contains about 3000 26pt Japanese characters and fits in 2048x2048, taking up 4MB. For Character Set, choose Characters from file and choose the txt file you added earlier. Leave Font Style at Normal/2 and change Font Render Mode to Distance Field 32 (Note: This is the slowest and highest quality. If you're still trying to figure out what font/atlas sizes you need, I recommend first choosing Hinted Smooth, which will slightly underestimate the size of the characters but be super fast, then use Distance Field 16 which will get all the character sizes right while still being faster than Distance Field 32, before finally choosing Distance Field 32 for the one you save).

Next, click Generate Font Atlas. This will take a while for large atlases. When it's done, it will say in the box if it was unable to fit any characters. If the resulting atlas is completely full, this is probably because those characters didn't fit, in which case you should adjust your atlas size or font size. Otherwise, this is because TextMeshPro failed to find those characters in your font. If it's just some newline characters or spaces this is fine, just look through them to make sure it didn't miss any characters that you think you need.

Once you've saved the font, use the Unity inspector to change its Line Height to be equal to its Point Size. This is what MS Gothic uses, and a lot of the game's character spacing expects this to be the case.

From here, follow the directions above for TMPAssetConverter to add your font to the game.