FrameShift

Transform your media in two clicks.

The free, open-source FFmpeg GUI for Windows. Convert, compress, crop, cut, extract audio, create subtitle files with local Whisper, upscale images and video, remove backgrounds, remove objects, split audio stems with Demucs, smooth motion with RIFE and more — straight from the Explorer right-click menu. 100% local. 100% offline. No upload, ever.

Download View on GitHub
FrameShift Explorer menu demo
  • New in 1.15.0: Create Subtitle File
  • Windows-first
  • Runs locally
  • Powered by FFmpeg

Why FrameShift?

Two clicks. That's it.

Right-click a file, pick an action, you're done. No app to open, no import, no export. The workflow disappears so you can focus on the file.

Everything stays on your machine.

No cloud, no upload, no telemetry. Your photos, videos and audio files never leave your disk. Works the same with or without an internet connection.

Built on FFmpeg.

The open-source reference for audio and video, natively integrated. Reliable, fast, and battle-tested — no proprietary black box.

Built for quick corrections.

Need to crop, trim, rotate, resize, or extract something fast? FrameShift is designed for small, repeated media fixes that should take seconds, not minutes.

No messy outputs.

Results stay next to the original file with safe unique naming. You keep your source, avoid accidental overwrites, and stay in control of every version.

Local AI, when it helps.

Use AI features like subtitle creation, image and video upscaling, background removal, denoising, and audio separation locally on your machine. No upload, no subscription wall, no sending private files to a remote service.

See it in action

Before: original image with object to remove Before
After: object removed by local AI inpainting After
AI · runs locally

Remove Object

Paint over any object with a brush, click Apply — the AI reconstructs the background behind it. No upload, no cloud. Runs entirely on your machine with LaMa inpainting via ONNX Runtime.

Upscale video demo
AI · runs locally

Upscale Video

New in FrameShift 1.14.0, the video upscaler brings low-resolution clips into the same local-first workflow as the rest of the app. Improve compressed or older footage directly from Explorer, with no upload and no cloud dependency.

Upscale image model picker
AI · runs locally

Upscale Image 4x

Enlarge an image x4 with a local AI model picker built for quick choices: Real-ESRGAN x4plus for general images, Anime 6B for illustration work, and Swin2SR for restoration-focused quality. DirectML runs when available, with automatic CPU fallback and invisible tiling for large files.

Remove background demo
AI · runs locally

Remove Background

Cut out subjects from your images with a single click. Choose between fast and higher-resolution local models, with optional BRIA user-supplied modes for non-commercial setups. No subscription, no upload.

Image to PDF demo

Image to PDF

Select your images, choose the action, get a clean PDF ready to share. Order, page size and orientation handled for you.

Interpolate video demo
AI · runs locally

Interpolate Video (RIFE)

Generate smoother motion with the local RIFE workflow, model preflight, and ONNX Runtime DirectML acceleration when available.

Local AI is now part of FrameShift.

Nine local AI features now run locally inside FrameShift through ONNX Runtime. Upscale Image 4x enlarges images locally with three selectable models for general use, illustration, or restoration work. FrameShift 1.14.0 adds Upscale Video to the same local-first toolbox for low-resolution clips that need a cleaner, larger result. FrameShift 1.15.0 adds Create Subtitle File, a local Whisper workflow that transcribes audio or video into an adjacent unique SRT file with three selectable models. Remove Object uses LaMa inpainting — paint a mask over any object and the AI fills it in. Remove Background uses a small local model catalog, including BiRefNet Lite plus higher-resolution modes. Remove Noise (Audio and Video) uses DeepFilterNet3 exports for offline speech denoising. Audio Separation is powered by HTDemucs exports with CPU and DirectML GPU support. Interpolate Video (RIFE) uses a local interpolation model with shared preflight and downloader flow. More local AI modules are planned for future versions, and they will follow the same rule: always local, never in the cloud.

See all 9 local AI features →

All actions, at a glance

Under the hood

  • Windows 10 / 11 · .NET 8 LTS · self-contained win-x64
  • FFmpeg & FFprobe bundled — no external install required
  • Clean Inno Setup installer, complete uninstall
  • Local AI via ONNX Runtime + DirectML (GPU when available, CPU fallback)
  • Open source · GPL v3.0

Frequently asked questions

Is FrameShift free?

Yes. Completely free, no subscription, no trial period, no feature lock. It is also open source under the GNU GPL v3.0 license.

Does it send my files anywhere?

No. All processing happens locally on your machine. Your photos, videos and audio files never leave your disk. No upload, no cloud, no telemetry of any kind.

Does it require FFmpeg to be installed separately?

No. FFmpeg and FFprobe are bundled with FrameShift. Nothing to install or configure manually — the installer handles everything.

Does it work without an internet connection?

Yes. FrameShift is fully offline. An internet connection is only used the first time you download an optional AI model (background removal, subtitle creation, image or video upscaling, etc.). After that, everything runs without any connection.

What Windows versions are supported?

Windows 10 and Windows 11 (x64). FrameShift is self-contained and requires no additional runtime to be installed.

How is it different from HandBrake or Shutter Encoder?

HandBrake and Shutter Encoder are excellent dedicated converters. FrameShift is complementary: it is designed for fast, everyday corrections launched directly from the Windows right-click menu without opening a separate app — and it includes local AI workflows such as background removal, image and video upscaling, audio stem separation, subtitle creation, denoising, and RIFE interpolation.

Does it use the GPU for AI tasks?

Yes, when available. FrameShift uses ONNX Runtime with DirectML to run AI models on the GPU. If no compatible GPU is found, it automatically falls back to CPU processing.

Can I remove the background from an image without uploading it?

Yes. The background removal feature runs entirely on your machine using a local AI model. No account, no upload, no subscription required.