Skip to content

NumberZeroo/TextTo3D-WebXR

Repository files navigation

TextTo3D-WebXR

This project enables the generation of 3D models from textual prompts. The pipeline first produces an image using FLUX, then constructs a 3D model with StableFast, and finally UniRig applies rigging to the output.

Requirements

  • Node.js ≥ 18
  • Python ≥ 3.10
  • Conda (for the rigging environment)
  • WebXR-compatible browser (Chrome, Edge, Oculus Browser, etc.)

Installation

git clone https://github.com/NumberZeroo/TextTo3D-WebXR.git
cd TextTo3D-WebXR

Backend (Flux + StableFast + UniRig)

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Then navigate into stablefast repo and install the requirements.

conda create -n unirig python=3.10 -y
conda activate unirig

Navigate into UniRig repo and install the requirements.

Frontend

There are two available frontend versions:

  1. Main frontend (included in this repository), developed with three.js.

This is the default frontend already present here.

npm install
vite
  1. Alternative frontend, developed with Meta Immersive Web SDK.

This version is located in a separate repository

git clone https://github.com/NumberZeroo/TextTo3D-FE-IWSDK-.git
cd TextTo3D-FE-IWSDK-
npm install
npm run dev

Usage

1. Enter VR/WebXR mode
2. Use the virtual keyboard to type a prompt
3. Wait for the pipeline (image → 3D model → rigging)
4. Enjoy!

About

This project enables the generation of 3D models from textual prompts.

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages