This is no longer maintained and I won't be responding to requests for updates/help/tips/etc, sorry!

Nintendo's Face Library is responsible for rendering Mii characters across numerous games and devices. This project's goal was to reverse-engineer it and provide utilities for extracting the assets that it uses.

Credits:

• Jaames - Python implementation and texture/model data
• Cholip - Python fixes
• Kinnay - reverse-engineering resource archive format

This utility can extract textures and 3D models from Face Library asset archives -- textures are converted to .png images and models are converted to binary glTF (.glb) models.

(Protip! You can import .glb files into Blender with this plugin).

Requirements:

• Python 3 (tested with 3.7.0)
• numpy
• Pillow

At the moment, Face Library asset archives from Miitomo can still be downloaded from archive.org:

• AFLResHigh.dat
• AFLResHigh_2_3.dat

Usage:

python3 fflExtract.py -i <face library archive> <tex count> <mesh count> -t <tex output dir> -m <mesh output dir>

tex count and mesh count depend on the file you're extracting:

• FFLResHigh, FFLResMiddle (Wii U): 317 textures and 857 meshes
• AFLResHigh (Miitomo):317 textures and 857 meshes
• AFLResHigh_2_3 (Miitomo): 365 textures and 859 meshes