|   | 

简体中文 | 繁體中文 | English | ไทย | Tiếng Việt | 日本語 | Русский

🌐 Multi-language Support

Toonflow supports the following interface languages:

Language	Language
Chinese (Simplified)	简体中文
Chinese (Traditional)	繁體中文
English	English
Thai	ไทย
Vietnamese	Tiếng Việt
Japanese	日本語
Russian	Русский

💡 More languages are being adapted — contributions for translations are welcome!

🌟 Key Features

Toonflow is an AI workstation designed for short drama production, building a complete closed loop around "Planning → Scriptwriting → Storyboarding → Final Output," supporting a localized, programmable, and continuously iterable production workflow.

• ✅ Infinite Canvas Production Workbench
  Organize scripts, characters, storyboards, assets, and video nodes in an infinite canvas-like layout, supporting free arrangement, backtracking, and parallel production without linear constraints.
• ✅ Three-layer Agent Collaboration System
  Decision, execution, and supervision layers work together, covering task decomposition, content generation, quality review, and revision feedback, improving stability and output consistency.
• ✅ Persistent Agent Memory
  Cross-session memory system based on local ONNX vector retrieval, supporting short-term messages, long-term summaries, and semantic recall, ensuring multi-round creative continuity.
• ✅ Programmable Provider System
  Write vendor TypeScript logic directly in the settings center, taking effect instantly without modifying source code or restarting, making it easy to privatize and integrate multiple models.
• ✅ Chapter Event Graph-driven Adaptation
  Automatically extract chapter events from original novels and store them structurally. Script adaptation uses event graphs to precisely invoke context, reducing information loss in long texts.
• ✅ Skill File Configuration
  Core prompts for ScriptAgent and ProductionAgent are externalized as Markdown Skill files, supporting online editing and rapid tuning.

📦 Application Scenarios

• Short video content creation
• Novel-to-film experimentation
• AI literary adaptation tool
• Script development and rapid prototyping
• Video asset generation

🔰 User Guide

Quick Start

1. Launch the application and log in (default: admin / admin123).
2. Complete model vendor configuration in the settings center (text/image/video models).
3. Create a new project and import the original novel, then execute chapter event extraction.
4. Enter ScriptAgent to generate the story skeleton, adaptation strategy, and structured script.
5. Switch to ProductionAgent to organize storyboards, assets, and video nodes on the infinite canvas.
6. Perform node-based refinement on storyboard images, then return them to the workbench for video stitching and export.

📺 Video Tutorial

https://www.bilibili.com/video/BV1oXD7BqEqJ

Toonflow 12-minute Quick Start AI Video 👉 Click to Watch

📱 Scan with WeChat to watch

📸 Demo Screenshots & Video

The following screenshots and video are from a short AI drama demo made with Toonflow, completed in about 2 hours, covering script generation, storyboard creation, and editing.

🎬 Demo Video

https://github.com/user-attachments/assets/2d9fddac-dfdf-4640-b030-b09d7f7287e9

If the video cannot play, please click to download

Demo Info

Item	Details
Production Cycle	Approximately 2 hours
Video Model	Seedance 2.0
Image Model	GPT Image 2
Language Model	Claude Opus 4.6
Total Duration	Approximately 2 minutes (raw 3 min, cut ~1 min of unusable footage)

Cost Breakdown

Model Type	Cost
Language Model	~￥10
Video Model (full generation)	~￥120
Image Model	Less than ￥1
Total	~￥130

Disclaimer: The demo's original resolution is 1080×1882; the release version has been compressed to 480p. If there are any copyright issues, please contact us to delete it.

🚀 Installation

Prerequisites

Before installing and using this software, please prepare the following:

• ✅ Large language model AI service API endpoint
• ✅ Sora or Doubao video service API endpoint
• ✅ Nano Banana Pro image generation model service API

Local Installation

1. Download & Install

OS	GitHub	Notes
Windows	Release	Official release package
Linux	Release	Official release package
macOS	Release	Official release package

[!CAUTION] On macOS, go to Settings > Privacy & Security to configure security settings; otherwise, the app may not open due to certificate issues.

2. Start the Service

After installation, launch the program to start using the service.

⚠️ First Login
Username: admin
Password: admin123

Docker Deployment

Prerequisites

• Docker installed (version 20.10+)

Method 1: Online Deployment

To be completed; use local build for now.

Method 2: Local Build

# Clone the project first (skip if already cloned)
git clone https://github.com/HBAI-Ltd/Toonflow-app.git
cd Toonflow-app

# Build and start locally using docker-compose
yarn docker:local

# Or build manually
docker build -t toonflow .
docker run -d -p <local_port>:10588 -v <local_data_path>:/app/data toonflow

Service Port Description

Port	Purpose	Deployment Mapping
10588	Software UI	10588:10588

🔧 Development Guide

[!CAUTION] 🚧 PR Submission Guidelines 🚧

⛔ Do not submit PRs to the master branch. ✅ Please submit PRs to the develop branch.

Developers are welcome to contribute to Toonflow. If interested, please contact the maintainer ACT in the community group.

🛠️ Tech Stack

Category	Technology
Runtime	Node.js 23.11.1+
Language	TypeScript 5.x
Backend	Express 5
Database	SQLite (better-sqlite3 / knex)
AI Integration	Vercel AI SDK (OpenAI / Anthropic / Google / DeepSeek / Zhipu / MiniMax / Tongyi Qianwen / xAI)
Local Inference	@huggingface/transformers (ONNX)
Real-time	Socket.IO
Desktop	Electron 40
Image Processing	Sharp
Containerization	Docker

Development Environment Setup

• Node.js: Version 23.11.1 or higher
• Yarn: Recommended as package manager

Quick Start Project

1. Clone the Project

git clone https://github.com/HBAI-Ltd/Toonflow-app.git
cd Toonflow-app
yarn install

2. Start Development Environment

yarn dev

Or start the Electron desktop client:

yarn dev:gui

Production mode:

yarn start

3. Build and Package

yarn build
yarn dist:win
yarn dist:mac
yarn dist:linux

4. Code Quality Check

yarn lint

🔗 Related Repositories

Repository	Description	GitHub	Gitee
Toonflow-app	Full client (this repo, recommended for regular users)	GitHub	Gitee
Toonflow-web	Frontend source code (for frontend developers)	GitHub	Gitee

💌 Contact Us

📧 Email: [ltlctools@outlook.com](mailto:ltlctools@outlook.com?subject=Toonflow Inquiry)

📜 License

Toonflow is open-sourced under the Apache-2.0 license, with supplementary commercial terms.

License details: https://www.apache.org/licenses/LICENSE-2.0

Supplementary Agreement

• If you distribute this software as a product to 2 or more independent third parties, you must obtain written commercial authorization from HBAI-Ltd.
• ≤ 5 legal entities jointly operating for internal use without providing services externally is considered internal use and does not require authorization.
• Do not remove or modify any Toonflow identifiers or copyright information.

Permanent Free Use Cases

• ✅ Creating content with Toonflow and receiving platform revenue shares
• ✅ Secondary development for internal team use
• ✅ ≤ 5 legal entities joint internal operation
• ✅ Personal learning, research, non-commercial use

Commercial License Pricing

Stage	Annual Revenue	Annual Fee
🌱 Support	< ¥100,000	Free license upon request
🚀 Startup	¥100–500,000	¥5,000/year
📈 Growth	¥500k–1.5M	¥20,000/year
🏢 Scale	¥1.5M–5M	¥80,000/year
🌐 Enterprise	> ¥5M	Negotiable

Non-retroactive clause: Users who used Toonflow under AGPL-3.0 before the v1.0.8 release will continue to be governed by AGPL-3.0 and are not affected by this agreement change.

See the full agreement in the LICENSE file.

🙏 Acknowledgements

We thank the open-source projects and organizations that provide support to Toonflow, including Express, AI SDK, Better-SQLite3, Sharp, Axios, Zod, Socket.IO, Electron, and Hugging Face Transformers.

For the complete list of third-party dependencies, please refer to NOTICES.txt.

copyright © Beijing Ai'a Technology Co., Ltd.