figura2vrc — Figura → VRChat avatar conversion pipeline

Converts a Figura (Blockbench) Minecraft player avatar into a working VRChat avatar: a rigid-limb Unity-Humanoid rig, PhysBones-ready skeleton, and in-game expression toggles built from the model's Blockbench animations.

Debugged end to end on real conversions (two avatars and counting) — the docs include every pitfall hit along the way so you don't have to rediscover them, plus a per-part PhysBones tuning table and a checklist for adapting the pipeline to a new avatar.

Quickstart

# 0. In Blockbench: File > Export > Export glTF  ->  model.gltf (with animations)

# 1. From your avatar folder (the one containing model.gltf):
blender --background --python path\to\figura2vrc\convert_to_vrchat.py
#    options after "--", e.g.:
#    ... convert_to_vrchat.py -- --height 1.6 --front-marker FrontHair --lift LeftBrow=0.5

# -> writes vrchat_export\ next to the glTF: avatar.fbx, textures, previews,
#    avatar_rigged.blend, convert_report.json

# 2. In your VRChat Unity project (Creator Companion, Avatars SDK):
#    copy avatar.fbx twice: Assets/avatar.fbx and Assets/avatar_anims.fbx,
#    plus the textures. Follow docs/CONVERSION.md for import settings.

# 3. Copy unity/FiguraExpressionSetup.cs into Assets/Editor/, adjust its
#    config block, then: Tools > Figura Avatar > Setup Expressions.

Full walkthrough — Unity import settings, humanoid mapping pitfalls, materials, PhysBones, publishing, and the troubleshooting table — in docs/CONVERSION.md.

Layout

Path	What
convert_to_vrchat.py	Blender headless converter: glTF → T-posed humanoid FBX
unity/FiguraExpressionSetup.cs	Unity editor script: FX animator, auto-blink, expressions menu
docs/CONVERSION.md	The full conversion guide + troubleshooting

Requirements

• Blockbench (glTF export with animations)
• Blender 4.2+ (developed on 5.2), used headless
• Unity 2022.3 via VRChat Creator Companion with the Avatars SDK
• The model must use the vanilla player bone names (Head, Body, LeftArm, RightArm, LeftLeg, RightLeg) — every standard Figura player avatar does.

Notes

• LIFT_PIXELS in the converter holds per-model cosmetic offsets (the original avatar needed its brow bars raised half a pixel clear of its glasses). It silently no-ops for models without those bones; override or clear it via --lift.
• Output is PC-targeted (Unlit/Transparent Cutout materials). For a Quest/Android build, re-shader with VRChat/Mobile/* equivalents.