What is this?
It's an AI Agent platform for controllable film and video production, from short clips to live-action style content, built as a creative workspace where you develop ideas with an AI Assistant, upload references, and generate and organize images and videos on a canvas.
It's an AI Agent platform for controllable film and video production, from short clips to live-action style content, built as a creative workspace where you develop ideas with an AI Assistant, upload references, and generate and organize images and videos on a canvas.Is this a hosted service, or do I run it myself?It's a self-hosted web application you deploy yourself with Docker; there's no hosted SaaS version described, and the recommended way to run it is the prebuilt Docker image distribution.
It's a self-hosted web application you deploy yourself with Docker; there's no hosted SaaS version described, and the recommended way to run it is the prebuilt Docker image distribution.How do I get it running?Paste the quick-start instructions into a local coding assistant with terminal access; it inspects your machine, installs dependencies, and configures Docker for you, and the release Compose setup starts Caddy and serves the app over HTTPS.
Paste the quick-start instructions into a local coding assistant with terminal access; it inspects your machine, installs dependencies, and configures Docker for you, and the release Compose setup starts Caddy and serves the app over HTTPS.Do I need Node.js, MySQL, or Redis installed first?No - the prebuilt Docker path runs Node.js, MySQL, Redis, Temporal, and FFmpeg inside containers, so Docker is all you need locally; you do need your own OpenRouter API key to actually generate images or videos.
No - the prebuilt Docker path runs Node.js, MySQL, Redis, Temporal, and FFmpeg inside containers, so Docker is all you need locally; you do need your own OpenRouter API key to actually generate images or videos.What's missing or limited in this preview?This is an early preview with rough edges; music and voiceover controls aren't included yet, and the MiniMax H3 Max model only accepts a single first frame instead of a first/last-frame pair.
This is an early preview with rough edges; music and voiceover controls aren't included yet, and the MiniMax H3 Max model only accepts a single first frame instead of a first/last-frame pair.Where do my project files and prompts end up?Project data and media stay in your own local database and private MinIO storage, but prompts and reference images are sent to the AI provider you configure through OpenRouter to actually run generation tasks.
Project data and media stay in your own local database and private MinIO storage, but prompts and reference images are sent to the AI provider you configure through OpenRouter to actually run generation tasks.What are the license terms?From v0.5.0-beta.1 the project is distributed under the Elastic License 2.0, a source-available (not OSI-approved) license that allows personal and internal business use but requires separate permission to resell it as a hosted or managed service.
From v0.5.0-beta.1 the project is distributed under the Elastic License 2.0, a source-available (not OSI-approved) license that allows personal and internal business use but requires separate permission to resell it as a hosted or managed service.Which AI models can I use for generation?The preview exposes OpenRouter configuration giving access to models like GPT Image 2, Nano Banana variants, Seedance variants, and MiniMax H3 / H3 Max; availability and cost depend on the provider you choose.
The preview exposes OpenRouter configuration giving access to models like GPT Image 2, Nano Banana variants, Seedance variants, and MiniMax H3 / H3 Max; availability and cost depend on the provider you choose.What languages does the interface support?The app UI itself currently supports Chinese and English; the Japanese and Korean versions are README translations only, not in-app language options.
The app UI itself currently supports Chinese and English; the Japanese and Korean versions are README translations only, not in-app language options.What can I do with a generated result?You can refine an existing image or video into a new version while keeping the original, so you can iterate on a result without losing earlier outputs.
You can refine an existing image or video into a new version while keeping the original, so you can iterate on a result without losing earlier outputs.What are similar projects?huobao-drama, printfilm, artcraft-services, dramaclaw, crm, ArcReel, Jellyfish, open-higgsfield
huobao-drama, printfilm, artcraft-services, dramaclaw, crm, ArcReel, Jellyfish, open-higgsfieldWhich topics is it in?storyboard, short-drama, prisma, s3, content-creation, tailwindcss, video-generation, mysql, redis, image-generation, nextjs, generative-ai
storyboard, short-drama, prisma, s3, content-creation, tailwindcss, video-generation, mysql, redis, image-generation, nextjs, generative-ai