Once Human (Beta) Noesis script

[TXKBOY]

2024年1月13日上午8点35分，DKDave 说：

是的，你先解压npk文件。你可以使用我的QuickBMS脚本：https://github.com/DKDave/Scripts/tree/master/QuickBMS

 

太酷了

Edited May 4 by TXKBOY

[TXKBOY]

I found a blog, he successfully unpacked, I hope it will be helpful to everyone    https://dumbrava.ca/posts/20241105-taking-apart-once-human-game-files/#cheating

[michalss]

11 hours ago, TXKBOY said:

I found a blog, he successfully unpacked, I hope it will be helpful to everyone    https://dumbrava.ca/posts/20241105-taking-apart-once-human-game-files/#cheating

ITs basically @DKDave work...

[DKDave]

9 minutes ago, michalss said:

ITs basically @DKDave work...

At least he left my name on it - some don't 🙂

 

[wq223]

On 8/18/2024 at 12:01 AM, Dabbius said:

Anyone been able to decompile the .pyc files? Im just trying to find the game data to make some online tools. Thinks like armor stats and rolls etc. Along with ui images for items. They have this thing locked down. I was able to unpack the .npk files and the script.npk is the juice with all the game stuff but i cant get the .pyc decompiled 

 

This game has recently been launched for the official version. The python version used seems to be very special and feels like a customized version because I haven't found this version in the public version
The version number corresponding to the magic number of pyc files exceeds the final version number of the official CPython 3.11 branch (3495)
Refer to the version definition in the CPython source code (importlib/_bootstrap_external.py), which has exceeded the standard release scope: https://github.com/python/cpython/blob/8303d32ff55945c5b38eeeaf1b1811dbcf8aa9be/Lib/importlib/_bootstrap_external.py#L421
I tried changing 3496 to 3495 and disassembly, and I could only process part of the content but it was much better than not being able to handle it before it was modified
For decompilation, they may need to reverse the game file to get the opcode, they use a custom opcode and need to map it to a standard opcode (if necessary, you should also find the actual version they are using)
By the way, pycdc support is not perfect, and many files are very confusing to decompile, so sometimes disassembly is also a good solution, at least for the text content is clearer.

[wq223]

9 hours ago, DKDave said:

At least he left my name on it - some don't 🙂

 

I cannot use the noesis script you wrote to load the official version of the .mesh file. Is it outdated?

mesh.zip

Edited May 5 by wq223

[shak-otay]

--

(see following post)

Edited May 5 by shak-otay

[DKDave]

6 hours ago, wq223 said:

I cannot use the noesis script you wrote to load the official version of the .mesh file. Is it outdated?

mesh.zip 1.01 MB · 3 downloads

Files look compressed so they won't work directly.  Not sure of compression method, although I think someone else posted a script recently to decompress other compression types.  Or upload a sample .npk archive so I can alter my script as well.

 

[wq223]

2 hours ago, DKDave said:

Files look compressed so they won't work directly.  Not sure of compression method, although I think someone else posted a script recently to decompress other compression types.  Or upload a sample .npk archive so I can alter my script as well.

 

The uploaded res_novice_pack_8.npk file can be processed using the bms script you wrote.
The script.npk file from the installation package cannot be processed. I found that it uses compression type 10, using the zstd header but cannot be decompressed (in all files, only this patch uses this compression type)
I have uploaded the file to cloud disk
This patch should allow for a closer test

https://drive.google.com/drive/folders/1R8ZnpQ8xZIzYyw8Qb6IytYWcoS8cLxUX?usp=sharing

 

Edited May 5 by wq223

[DKDave]

1 hour ago, wq223 said:

The uploaded res_novice_pack_8.npk file can be processed using the bms script you wrote.
The script.npk file from the installation package cannot be processed. I found that it uses compression type 10, using the zstd header but cannot be decompressed (in all files, only this patch uses this compression type)
I have uploaded the file to cloud disk
This patch should allow for a closer test

https://drive.google.com/drive/folders/1R8ZnpQ8xZIzYyw8Qb6IytYWcoS8cLxUX?usp=sharing

 

While type 3 is normal ZSTD which decompresses fine, types 4 and 10 seem to have the ZSTD 4-byte identifier, but then some modifications to the data so that it doesn't decompress normally.  The meshes that can be decompressed do have some slight differences to the ones in the beta, so I'll need to look at them sometime.

 

[12121212j]

Thank you for your efforts. Unfortunately, the latest version we downloaded still cannot read the mesh file.

 

I'll provide an NPK file for your reference.

res_normal_pack_28.7z

[12121212j]

20 hours ago, 12121212j said:

Thank you for your efforts. Unfortunately, the latest version we downloaded still cannot read the mesh file.

 

I'll provide an NPK file for your reference.

res_normal_pack_28.7z 106.35 MB · 4 downloads

I tried using GPT to remove references to .mtg, .mtl, and .pvr files, and after that, about 30% of the model could be read. I'm not sure where the issue lies, but this suggests the script still works — perhaps the model now uses a new data layout?