CineGen‑AI: Open‑Source AI Manga & Video Generator
CineGen‑AI: Open‑Source AI Manga & Video Generator
If you’re a comic creator, animator, or just a tech‑curious storyteller, you’ve probably seen how time‑consuming it can be to turn a single line of prose into a polished animated sequence. CineGen‑AI (available on GitHub at Will‑Water/CineGen‑AI) tackles this challenge head‑on, providing a keyframe‑driven workflow that keeps the narrative tight while keeping production fast.
What is CineGen‑AI?
CineGen‑AI is an end‑to‑end AI production studio built in TypeScript/React, designed to create motion comics, anime, and short dramas without a dedicated backend. The platform fuses:
- Google Gemini 2.5 Flash for high‑quality script parsing, visual prompt generation, and image‑to‑image inference.
- Veo 3.1‑Fast‑Generate‑Preview for instant keyframe interpolation, turning static images into smooth video.
- IndexedDB for local storage, ensuring privacy and zero‑cost deployment.
The result is an industrial‑grade tool that lets users move directly from a novel draft to a near‑final MP4 file.
Core Features
| Phase | Description | Key Tools |
|---|---|---|
| 1 – Script & Storyboard | Input a story outline; the model auto‑splits it into scenes, assigns moods, and outputs a storyboard‑ready prompt. | Gemini‑flash (text) |
| 2 – Assets & Casting | Generates character reference sheets and set concept art, with wardrobe variants and consistent facial features. | Gemini‑flash‑image |
| 3 – Director Workbench | Users design a grid of shots, specify Start and End keyframes, and rely on context‑aware generation to avoid continuity errors. | Veo‑interpolate |
| 4 – Export | Real‑time preview, progress tracking, and export of high‑res keyframes along with MP4 clip bundles ready for Premiere or After Effects. | Built‑in exporters |
Why Keyframe‑Driven?
Traditional text‑to‑video (TTV) models struggle with scene continuity and shot order. CineGen‑AI forces the user to draw first, animate later:
- Draw – Generate the exact start and end frames for each shot. These images are locked‑in and respect character and set consistency.
- Animate – Veo interpolates between the two frames, producing a video that preserves the composition and lighting of each keyframe.
The approach yields natural‑looking motion comics where each frame feels intentional and visually coherent.
Quick Setup Guide
- Clone the Repo
git clone https://github.com/Will-Water/CineGen-AI.git cd CineGen-AI - Install Dependencies
npm install - Configure API Key – Create a
.envfile in the root or setGM_API_KEYin your shell.echo "GM_API_KEY=YOUR_GEMINI_API_KEY" > .env - Run the Application
npm run dev - Access the UI – Open
http://localhost:5173in your browser.
Once the app is running, you can start drafting a story in Phase 1, generate assets in Phase 2, and walk through the storyboard in Phase 3. When you’re happy with the sequence, hit Export to get your MP4 and PNG assets.
Use Cases
- Independent Comic Creators – Push a finished story to platforms like Instagram or Webtoon faster.
- Animation Students – Prototype scenes and test camera angles before investing in full animation.
- Educational Content Creators – Produce engaging visual stories for science or history topics.
- Demo Reps and Pitch Decks – Quickly generate a short animated demonstration of a product or concept.
Extending CineGen‑AI
Because it’s open source, the community can contribute:
- Integrate additional model backends (e.g., Stable Diffusion for different artistic styles).
- Build new plugins for advanced camera rigs or motion tracking.
- Contribute localization files for more languages.
If you find a bug or have an idea, open an issue on GitHub – the project welcomes contributors!
Final Thoughts
CineGen‑AI is more than a demo; it’s a fully functional, production‑ready AI studio that puts control back in the hands of storytellers. With a focus on consistency, speed, and privacy, it opens a new frontier for comic and animation creation.
Try it today, share your creations, and help shape the future of AI‑powered storytelling.
Resources | Resource | Link | |----------|------| | GitHub repo | https://github.com/Will-Water/CineGen-AI | | Demo site | https://aniktku.com (under construction) | | Documentation | Readme files in multiple languages (en, zh, ja, ko) |
CineGen‑AI: Built for creators, by creators.