HubLensAIHBAI-Ltd/Toonflow-app
HBAI-Ltd

Toonflow-app

AI#AI#TypeScript#Electron#Generative AI#Video Production
View on GitHub
7,423

// summary

Toonflow-app is an AI workbench designed for short drama production, achieving full-process automation from script to video through an infinite canvas and a three-layer Agent collaboration system. The platform supports chapter event graph-driven adaptation and provides a programmable provider system to flexibly integrate various AI models. Users can leverage its persistent memory system and modular skill configuration to significantly improve the efficiency and consistency of short drama creation.

// technical analysis

Toonflow is an AI workbench designed specifically for short drama production, adopting a closed-loop architecture of "planning-scriptwriting-storyboarding-production," aiming to efficiently convert novel content into video episodes through AI automation technology. The project solves the problems of cumbersome traditional short drama production processes and difficult-to-control consistency through an infinite canvas workbench and a three-layer Agent collaboration system (decision-making, execution, supervision). In terms of technical decisions, the project adopts a programmable provider system and externalized Markdown Skill files, allowing developers to flexibly adjust AI logic without modifying the source code, reflecting high scalability and friendliness to private deployment.

// key highlights

01
The infinite canvas workbench supports the free arrangement of scripts, characters, and storyboards, breaking the limitations of traditional linear creation.
02
The three-layer Agent collaboration system significantly improves the stability and consistency of AI-generated content through task decomposition and quality supervision.
03
The persistent memory system based on local ONNX ensures context continuity during multi-round creation processes.
04
The programmable provider system allows users to write logic via TypeScript, enabling private model access without restarting.
05
The chapter event graph-driven adaptation function can accurately extract the structure of the original work, effectively reducing information loss in long-text processing.
06
Core prompts are externalized as Markdown Skill files, supporting rapid online tuning and lowering the management threshold for complex prompts.

// use cases

01
Novel-to-film adaptation and script development
02
Short video content creation and storyboard design
03
AI video asset generation and automated production

// getting started

Users can download the installation package for their corresponding operating system from GitHub or Gitee, and log in using the default account admin/admin123 after startup. After entering the settings center to configure the AI model provider interface, you can create a new project and import the original work, generate a script via ScriptAgent, and complete storyboarding and video production on the infinite canvas of ProductionAgent.