shuyu-labs/BigBanana-AI-Director

BigBanana AI Director是一个工业级一站式 AI 短剧，AI 漫剧,AI 导演平台，面向创作者，实现从灵感到成片的高效生产。 它摇弃了传统的“抽卡式”生成，采用 "Script-to-Asset-to-Keyframe" 的工业化工作流。实现 “一句话生成完整短剧，从剧本到成片全自动化”，同时精准控制角色一致性、场景连续性与镜头运动。

What it solves

BigBanana AI Director addresses the critical problem of character consistency in AI-generated motion comics and short dramas. Unlike "lottery-style" generation where characters change appearance between shots, this platform provides an industrial workflow to ensure that characters, scenes, and props remain visually stable across an entire series.

How it works

The platform uses a "Script-to-Asset-to-Keyframe" workflow rather than relying solely on text prompts. It breaks the production process into five phases:

  1. Narrative Planning: Converts story outlines or novels into structured scripts with defined characters, scenes, and shots.
  2. Consistency Assets: Creates "character look-books" (standard reference images) and a wardrobe system to lock in visual identities before filming begins.
  3. Shot Workbench: Uses a keyframe-driven approach where creators generate specific start and end frames, and the AI interpolates the motion between them.
  4. Delivery Center: Provides a timeline editor for rough cutting and exporting final video files.
  5. Prompt Management: Centralizes all prompts for characters, scenes, and assets to allow for global corrections.

Who it’s for

It is designed for serialized content creators, web-novel IP adaptation teams, and content studios who require high control over their visual assets and are willing to perform light manual fine-tuning rather than seeking one-click automation.

Highlights

  • Triple Consistency Guarantee: Combines standard reference photos, a wardrobe system for different outfits, and context-aware generation to prevent character "face-swapping."
  • Keyframe-Driven Control: Allows precise control over camera movement and shot transitions by defining start and end frames.
  • Local-First Data: Stores project data in the browser's IndexedDB to ensure creator privacy and IP security.
  • Multi-Platform Deployment: Available as a web app, a Docker-hosted self-hosted instance, or a desktop client for Windows and macOS.

Related

  • Project
    liangdabiao/Seedance2-Storyboard-Generator

    A structured AI video production workflow that helps creators transform stories into multi-episode video series using Seedance 2.0 and AI agent automation.

  • Project
    liyue-aigc/female-outfit-director

    A prompt-directing workflow for AI agents that generates structured production plans and prompts for creating character-consistent outfit-change videos.

  • Project
    zenstory-ai/drama-skills

    Drama Skills is an open‑source, agent‑compatible toolkit that lets writers turn a novel or idea into a finished short‑drama video. It provides eleven “skills” (commands) that cover the whole pipeline—analysis, script writing, asset extraction, image/video prompt generation, storyboard, production, editing, and review—using plain Markdown files as the single source of truth. Install via ClawHub, run the commands inside Claude Code, Codex, or any Agent‑Skill environment, confirm media generation before spending, and export a complete, auditable project.

  • Project
    lixiaoxiao9888-create/manju-laoli-skill

    A skill package for AI agents that automates short‑drama creation: lock model/aspect‑ratio, generate script, dialogue, assets, storyboard, and produce a ready‑to‑feed 7‑section prompt (or MiniMax H3 format). Includes validation scripts, offline storyboard board, and FACS‑based micro‑expression tools for consistent AI video production.

  • Project
    Forget-C/Jellyfish

    An end-to-end AI production workspace for short dramas that manages the full pipeline from script breakdown and asset consistency to video generation.