visualbruno/3DGenStudio

Orchestrate complete 3D generation pipelines — from text-to-image, image editing, mesh generation, UV unwrapping, to texturing — all in a single visual workspace powered by ComfyUI and external APIs.

🎨 What is 3D Gen Studio?

3D Gen Studio is an open‑source desktop/web application that lets you build complete 3‑D assets — from a text prompt all the way to a game‑ready mesh — inside a single visual workspace. It glues together ComfyUI (a node‑based AI image generation engine) and a handful of external 3‑D generation APIs, then adds a suite of mesh‑editing tools (UV unwrap, retopology, rigging, texture baking, LOD creation, etc.).

The core idea is a pipeline UI that can be displayed as either a Kanban board (cards flow through stages) or a node graph (assets are nodes linked by arrows). Each card or node can trigger a ComfyUI workflow or an external REST/GraphQL API, and the results automatically feed the next stage.

🛠️ Main Features (as described in the README)

Feature	What it does
Kanban board	Drag‑and‑drop cards through stages: Images → Image Edit → Mesh Gen → Mesh Edit → Texturing. Each card can store custom metadata and launch a ComfyUI workflow or API call.
Node‑graph view	Visual graph of assets; click a node to edit parameters, run the associated ComfyUI workflow, or view the mesh preview.
Assets library	Centralised view of images, meshes (GLB/OBJ/FBX), and ComfyUI workflow files. Supports versioning, format badges, and one‑click import into any project.
ComfyUI integration	Runs any ComfyUI workflow locally; parameters can be edited inline, and outputs are captured as new assets.
External API support	Any REST/GraphQL 3‑D service (e.g., TripoSR, Wonder3D, Kimodo) can be called from a card or node.
Local‑first storage	Projects are saved on disk (SQLite/LowDB), making them Git‑syncable and portable with no cloud lock‑in.
Mesh editor	In‑app tools for:

Texturing (project textures onto the mesh)
Basic modeling (face/vertex edits)
Sculpting (standard brushes)
Painting directly on the mesh
Displacement via brush‑driven tessellation
Projection of textures using ComfyUI
Auto‑UV unwrap, Auto‑Retopo, Auto‑Rig (AI‑driven skeleton/weights via SkinTokens & Kimodo)
Optimization/LOD generation, collision hull creation
Baking of normal/AO/position maps
Game‑ready checks (triangle budget, UV seams, scale, pivot) | | Batch projects | Define a series of stages that run automatically; supports image, mesh, and boolean types, and can export asset lists. | | Multi‑user / LAN | Server can be hosted on another machine; desktop app can connect over LAN or behind a reverse proxy. | | Cross‑platform desktop app | Built with Electron/Node, runs on Windows, macOS (Apple Silicon only), and Linux. | | Extensive changelog | Frequent updates (multiple per week) adding new generators, VFX tools, tree/assembly generators, animation integration (MoCapAnythingV2, Kimodo), and UI refinements. |

🚀 Typical Workflow

Create / import an image – use a ComfyUI text‑to‑image model or an external API.
Edit the image – in‑painting, normal‑map generation, background removal.
Generate a mesh – call a 3‑D generation service (e.g., TripoSR) from the card.
UV & texture – run auto‑UV, projection, or texture‑painting steps.
Finalize – bake maps, create LODs, rig the mesh, and export as GLB/OBJ/FBX.

All steps can be visualised as cards moving across the Kanban board or as linked nodes in the graph view.

📦 Installation (quick summary)

Prerequisites: Node .js + npm, a running ComfyUI instance, Python 3 for the backend server.
Clone repo → npm install → npm run dev for the frontend.
In python-server folder: create a virtualenv, pip install -r requirements.txt, then python main.py.
Backend runs on http://localhost:3001; Vite dev server serves the UI.
macOS Apple‑Silicon builds need a manual Gatekeeper bypass (see README).

🧩 Tech Stack

Frontend: React, Vite, Three.js (@react-three/fiber, @react-three/drei).
Backend: Node/Express + Multer for file uploads.
Storage: SQLite + LowDB for asset metadata.
AI integration: ComfyUI (local) + arbitrary external AI/3‑D APIs.

🤝 Community & Contributions

Open to PRs (badge in README).
Discord channel for support and feature discussion.
Wiki bundled with the app for detailed documentation.

📚 Bottom line

3D Gen Studio is a genuine, actively maintained open‑source project that stitches together AI‑driven image generation, 3‑D mesh creation, and a full suite of mesh‑editing tools into a unified visual pipeline. It targets artists, game developers, and researchers who want a single place to prototype AI‑generated 3‑D assets without being locked into a cloud service.