Skip to content
| Marketplace
Sign in
Visual Studio Code>Machine Learning>Frame Minion: AI Video GeneratorNew to Visual Studio Code? Get it now.
Frame Minion: AI Video Generator

Frame Minion: AI Video Generator

Preview

Okey Landers

|
2 installs
| (0) | Free
AI video generation in VS Code: create images, single-shot videos, and orchestrated music videos with OpenRouter and fal.ai.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Frame Minion - AI video, where you make everything else

Frame Minion: AI Video Generator

Frame Minion brings AI image and video generation into VS Code. Generate reference frames, turn them into clips, or hand a song to the Music Video Orchestrator and build a multi-segment music video without leaving your workspace.

Use models from OpenRouter's image and video catalog - Veo, Sora, Kling, Wan, Seedance, Hailuo, Gemini, GPT Image, FLUX, and more - with optional fal.ai support for specialist video workflows and storage.

Your prompts, projects, outputs, and provider keys stay local. Frame Minion is a tool that runs inside VS Code, not a hosted Frame Minion service.

Visit frameminion.video | Setup guide | Privacy guide | Video Prompting Field Guide

What You Can Make

Image creation sidebar

Generate images in an iterative, chat-style workspace. Upload references, control aspect ratio, switch OpenRouter image models, and save or reuse any generated frame.

Frame Minion image creation sidebar

Use the image sidebar as a frame factory:

  • Generate style, character, location, start-frame, and end-frame images.
  • Push a generated image directly into the Video tab as a start frame, end frame, or reference.
  • Send images to the Music Video Orchestrator as segment frames or project-level style / character / location references.
  • Keep iterating while your past prompts and generated images remain visible.

Video creation sidebar

Create single-shot AI video from a focused render form. Pick a model, write the action/camera prompt, attach optional reference media, and render.

Frame Minion video creation sidebar

The video sidebar supports:

  • Text-to-video and image-to-video workflows.
  • Start frame, end frame, reference image, and audio inputs where supported by the selected model.
  • Model-aware duration, resolution, and aspect-ratio controls.
  • Saved original video files, webview previews, and quick reveal actions.
  • Continuation workflows that reuse the saved original clip correctly.

Music Video Orchestrator

The Music Video Orchestrator turns a song and concept into a structured, multi-segment video project. It can plan the song, generate start/end frames for each segment, produce per-segment clips, preserve alternate takes, add transitions, and render the final cut with audio.

Music Video Orchestrator storyboard

Use it for:

  • Song-aware planning with sections, timecodes, mood, lyrics, and segment prompts.
  • Project style references plus named character and location references.
  • Per-segment start frames, end frames, video prompts, and generated clips.
  • Version history for frames and videos, so alternate takes are not lost.
  • Transition presets, timeline sync warnings, preview render, and final render.

The Music Video Orchestrator is live today. More orchestrators are coming soon for other video workflows, including narrator, cutting-room, vertical, and trailer-style flows.

Frame Minion orchestrators tab

Guides

Frame Minion's public guides live at frameminion.video/guides.

  • Providers & Setup explains OpenRouter, fal.ai, storage, and ffmpeg.
  • Privacy & Data explains what stays local, what is sent to configured providers, and how keys are stored.
  • Video Prompting Field Guide collects the film vocabulary video models actually honor, with copy-ready phrasing.

Setup

  1. Install Frame Minion from the VS Code Marketplace.
  2. Open the Frame Minion view from the activity bar.
  3. Open Settings from the gear icon.
  4. Add your OpenRouter API key.
  5. Add a fal.ai key if you plan to generate video with start frames, end frames, reference images, or audio.
  6. Optional: install or let Frame Minion download ffmpeg when you render multi-clip projects.

Image generation and text/audio analysis only need an OpenRouter key. Video generation is different: video models need URLs for start frames, end frames, reference images, and audio instead of receiving those files inline with the prompt. Frame Minion uses fal.ai's free-with-account file hosting by default to create short-lived, obscured URLs the models can fetch during generation. Power users can use their own S3 bucket instead, but fal.ai is the intended zero-setup path.

Free To Build, Licensed To Final Render

Frame Minion is bring-your-own-key: you pay OpenRouter, fal.ai, or your storage provider directly for the model work they perform.

Most of Frame Minion is free to use:

  • Image generation.
  • Single-shot video generation.
  • Music Video Orchestrator planning.
  • Audio analysis and prompt helpers.
  • Frame generation and per-segment video generation.
  • Preview renders of a full orchestrated music video.

The paid Frame Minion license unlocks the Music Video Orchestrator's full-resolution Final render. You can buy or activate a license from Settings or from the locked Final button. Preview stays available without a license, so you can build and review the whole project before deciding to export the final file.

Provider And Model Support

Frame Minion is built around bring-your-own-provider accounts:

  • OpenRouter powers planning, prompt enhancement, image generation, audio analysis, and most video generation requests.
  • fal.ai provides the default short-lived hosting URLs that video models need to retrieve start/end frames, reference images, and audio. It also supports direct Wan-family workflows.
  • S3 can replace fal storage if you want media uploaded to infrastructure you control and are comfortable creating credentials and a bucket yourself.
  • ffmpeg handles local preview/final rendering for orchestrated projects.

Supported model families include:

  • Video: Veo, Sora, Kling, Wan, Seedance, Hailuo, and other OpenRouter video models.
  • Image: Gemini, GPT Image, FLUX, and other OpenRouter image models.
  • Text / multimodal: the configured OpenRouter planning and enhancement model.

Privacy

Frame Minion runs inside VS Code. There is no Frame Minion account and no hosted Frame Minion generation service.

  • API keys are stored in VS Code secret storage.
  • Prompts and media are sent directly to the providers you configure.
  • Project files and generated outputs are saved locally in your workspace or configured output directory.
  • The extension does not need your repository to be public.

Read the full Privacy & Data guide for provider-specific details and tradeoffs.

Configuration

Setting Default Description
frameMinion.imageModel google/gemini-3.1-flash-image-preview Image generation model ID
frameMinion.videoModel google/veo-3.1-fast Video generation model ID
frameMinion.openRouterModel openai/gpt-5.1 Text / multimodal model used for planning and prompt enhancement
frameMinion.outputDirectory frame-minion Workspace-relative folder for saved generations and projects
frameMinion.storageProvider fal Upload provider for reference media used by video models: fal or s3
frameMinion.maxConcurrentVideos 1 Maximum per-segment video generations to run in parallel in the Music Video Orchestrator
frameMinion.ffmpegPath empty Optional absolute path to a custom ffmpeg binary

Learn More

Frame Minion animated logo

Start with the Frame Minion website, then keep the guides close while setting up providers and shaping prompts.

Questions, bugs, or feature requests? Email support@frameminion.video.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft