Super 3D Pro - A practical Image-to-3D WebUI

Hi everyone,

I’m excited to share Super 3D Pro – an open‑source, self‑hostable web application that turns a single image or a text prompt into a high‑quality 3D model.

You can run it fully locally (using TripoSR + baking) or use cloud APIs (Tripo + Agnes AI) for even higher quality. The UI is built with Three.js so you can instantly preview, orbit, zoom, and export your model.

🔗 GitHub repo: [Insert your GitHub link here]
💡 Live demo (optional): [Add if you have one]

✨ Key features

• Two input modes

  • 🖼️ Image → 3D (upload any photo, background removed automatically)

  • 📝 Text → 3D (powered by Agnes AI image generation → Tripo)

• Two generation engines

  • 🖥️ Local (TripoSR) – fast, runs on GPU/CPU, no API key needed

  • ☁️ Cloud (Tripo API) – higher detail, PBR materials, ideal for AAA‑like assets

• Interactive 3D viewer

  • Drag to rotate / pan / zoom

  • Real‑time GLTF/GLB preview

• Export formats – .glb or .obj (as ZIP with textures)

• Batch processing – upload multiple images, generate one model after another

• Rating system (RLHF) – help improve future models by rating results

• Completely open‑source – no hidden telemetry, no paywalls

🧠 Under the hood

• Backend: FastAPI + asyncio + background tasks

• 3D viewer: Three.js (WebGL) with Draco & GLTF support

• Background removal: rembg + ISNet (local)

• Local 3D reconstruction: TripoSR (from Stability AI) with texture baking

• Cloud option: Tripo API (tsk_ keys) + Agnes AI for text‑to‑image

All processing happens on your machine (when using local mode) – no data leaves your computer unless you explicitly use cloud APIs.

🚀 Getting started (5 minutes)

bash

git clone https://github.com/yourusername/...
cd super3d-pro

# Install dependencies
pip install -r requirements.txt   # (create it, or install torch, fastapi, rembg, trimesh, etc.)

# Run the server
python main.py

Then open your browser to http://127.0.0.1:8888 – that’s it.

Note: for local mode you need PyTorch with CUDA (if you have a GPU). Cloud mode only needs an API key.

📸 Example workflow (local)

1. Upload a photo of a character, object, or toy

2. Optional: adjust output resolution (256–512)

3. Click Generate – wait 20–40 seconds

4. Preview the 3D model in the viewer

5. Export as .glb or .obj (with textures)

Text mode works similarly: enter a prompt + your Agnes API key → it generates an image first, then the 3D model.

❤️ Why open source?

I built this because I wanted a clean, modern, one‑click UI for 3D generation that I could run locally or extend easily.
Most existing solutions are either closed source, require complex command lines, or are tied to a single cloud service.

Super 3D Pro is my attempt to give back to the community – free, transparent, and hackable.

If you find it useful, please ⭐ star the repo – it helps others discover it.
And if you’d like to support my work, there’s a PayPal link inside the app (or you can sponsor me via GitHub).

📢 Call to action

• Try it, break it, report issues

• Add your own engine (Hunyuan3D, Stable Zero123, …) – the pipeline is modular

• Share your generated 3D models with #Super3DPro