Concept-2-3D

Concept-2-3D is a full-stack 3D generation and discovery project with:

• A FastAPI backend for intent parsing, model search, AI chat, reviews, and image-to-3D generation.
• A React + Vite frontend for searching, viewing, and interacting with 3D assets.
• ML support files for TripoSR-based image-to-3D workflows.

Monorepo layout

• Backend: FastAPI service and APIs.
• Frontend: React app (Vite).
• ML/core: TripoSR helpers, shims, and model-weight utilities.
• scripts: one-command virtual environment setup for Windows and Linux/macOS.

Features

• Intent analysis endpoint for prompt understanding.
• Multi-source model search with cache support.
• AI chat endpoint with provider fallback behavior.
• Review and rating APIs for models.
• Image-to-3D generation endpoints:
  • Synchronous: api/generate_from_image
  • Asynchronous with polling: api/generate_from_image_async and api/generate_status/{job_id}

Prerequisites

• Python 3.10+ recommended
• Node.js 18+ and npm
• Git

Quick start

1) Clone and enter project

git clone https://github.com/GuruMachanica/Concept-2-3D.git
cd Concept-2-3D

2) Create a single Python virtual environment

Windows PowerShell:

.\scripts\setup_venv.ps1

Linux/macOS:

./scripts/setup_venv.sh

3) Configure backend environment

Copy Backend/.env.example to Backend/.env and update values as needed.

At minimum, configure the providers you plan to use (for example GROQ or FREE_AI settings).

4) Install frontend dependencies

cd Frontend
npm install
cd ../..

5) Run backend

From repository root:

.venv\Scripts\python.exe -m uvicorn Backend.main:app --host 0.0.0.0 --port 8011 --reload

Why port 8011:

• Frontend Vite proxy defaults to http://localhost:8011 for both /api and /models.

6) Run frontend

In a second terminal:

cd Frontend
npm run dev

Open the local Vite URL shown in terminal (usually http://localhost:5173).

Backend API overview

Common endpoints:

• POST api/intent
• POST api/search
• POST api/chat
• POST api/cache/clear
• POST api/generate_from_image
• POST api/generate_from_image_async
• GET api/generate_status/{job_id}
• POST api/reviews/submit
• GET api/reviews/{model_id}
• GET api/reviews/{model_id}/summary
• GET api/reviews/{model_id}/user/{user_id}
• POST api/labels/position-from-image

When backend is running locally on port 8011, docs are available at:

• http://127.0.0.1:8011/docs

Notes on image-to-3D

• Runtime outputs are written under Backend/models.
• Large local model-weight artifacts should stay out of git.
• TripoSR generation failures are surfaced through async status error fields and latest run logs.

Development tips

• Lint backend (if ruff is installed in venv):

.venv\Scripts\python.exe -m ruff check Backend

• Lint frontend:

cd Frontend
npm run lint

Troubleshooting

• Frontend cannot reach backend:
  • Ensure backend is running on port 8011, or set BACKEND environment variable for Vite proxy.
• Image-to-3D fails immediately:
  • Check Backend/models latest tripo run folder and run.log.
• AI chat returns provider errors:
  • Verify GROQ or FREE_AI environment configuration in Backend/.env.

Related docs

• README_SETUP_VENV.md for venv setup details.

License

• This project is distributed under the Inspiration-Only License. See LICENSE.