Summary

ArcReel is an open-source AI-powered video generation workspace automating short video creation from novels. It targets users seeking an end-to-end solution, offering AI agent-driven scripting, character/scene consistency, and multi-vendor AI model integration for streamlined video production.

How It Works

A multi-agent architecture, using Claude Agent SDK, orchestrates a pipeline from novel analysis to video synthesis. It leverages "Skills" and "Subagents" for automated tasks. Key advantages include multi-vendor support (Gemini, Volcengine Ark, Grok, OpenAI) for image, video, and text generation, enabling flexibility and cost optimization. Features like style reference images and clue tracking ensure visual consistency across generated assets and shots, addressing a core challenge in AI video production.

Quick Start & Requirements

• Default Deployment: Clone the repository, navigate to /deploy, copy .env.example to .env, and run docker compose up -d. Access the UI at http://localhost:1241.
• Production Deployment: Navigate to /deploy/production, copy .env.example to .env (setting POSTGRES_PASSWORD), and run docker compose up -d.
• Initial Configuration: After first launch, log in and configure API keys for Anthropic (AI Assistant) and at least one AI provider (Gemini, Volcengine Ark, Grok, OpenAI) via the /settings page.
• Prerequisites: Docker, Python 3.12+, Node.js 20+, uv, pnpm, ffmpeg (for local development).
• Documentation: Links to a full getting started tutorial, CapCut export guide, and cost explanations are available.

Highlighted Details

• End-to-End Pipeline: Automates novel-to-video creation, including scriptwriting, character design, storyboarding, and video generation.
• Multi-Vendor AI: Supports Gemini, Volcengine Ark, Grok, OpenAI for text, image, video, with global and project-level switching.
• Custom Providers: Integrates any OpenAI or Google-compatible API endpoints (e.g., Ollama, vLLM).
• Visual Consistency: Ensures character and prop/scene continuity via AI design references and clue tracking.
• Content Modes: Offers "narration" (rhythm-based) and "drama" (scene/dialogue-based) structuring.
• Export Options: Generates CapCut (Jianying) draft files for further editing and supports project import/export.
• Cost Tracking: Monitors expenses across different AI vendors.

Maintenance & Community

• The project welcomes contributions via code, bug reports, and feature suggestions.
• A Feishu (Lark) group is available for community support and updates.

Licensing & Compatibility

• License: AGPL-3.0.
• Compatibility: AGPL-3.0's strong copyleft requires derivative works to be open-sourced under the same license, potentially restricting integration with proprietary closed-source applications.

Limitations & Caveats

• The AGPL-3.0 license's copyleft nature requires careful consideration for commercial or closed-source integration.
• Setup involves obtaining and configuring multiple AI service API keys, which may incur costs and complexity.