Convert video locally, with the codec choices made visible

Implemented output pipelines

• WebM: the browser decodes playback, frames are drawn to a Canvas at 30 fps, and MediaRecorder writes VP8 or VP9 where supported.
• MP4: FFmpeg encodes H.264 video with AAC audio and web-friendly fast-start metadata.
• MKV: FFmpeg encodes VPx video and Opus audio.
• MOV and M4V: FFmpeg uses MPEG-4 video and AAC audio.

Compression and remuxing

With compression enabled, video is resized inside the selected maximum dimensions and re-encoded at one of three video bitrates. The default is a 1280 × 720 bounding box. Re-encoding is lossy and CPU-intensive. With compression off, a same-format input is returned unchanged; a different output container first attempts to copy all streams without re-encoding. Stream copy is fast and lossless when the target container accepts every source stream, but it can fail when codecs are incompatible.

Tracks and metadata

The interface does not provide selectors for audio tracks, subtitles, chapters, or attachments. Stream-copy mode maps all streams, but the destination container may reject them. Re-encoding arguments do not promise preservation of every track or subtitle. Check the result before deleting a multi-track original.

Performance and browser limits

FFmpeg runs as WebAssembly inside the browser—not as a native desktop process. The source, decoded frames, encoder state, and virtual input/output files may coexist in memory. A 30 MB source can therefore require substantially more than 30 MB of working memory. Conversion can heat a phone, drain its battery, or fail if the tab is suspended. The local ZIP removes the website's artificial size check but retains these browser constraints.

Input compatibility

The picker accepts MP4, MOV, MKV, WebM, M4V, and files the browser identifies as video. FFmpeg handles inputs for its output paths, while WebM output depends more heavily on native browser playback support. An accepted extension is not a guarantee that every codec variant will decode.