Video Compressor

Shrink videos fast with hardware acceleration: pick High, Basic or Low compression with estimated final sizes, or an exact target size.

  • Runs in your browser
  • Free, no sign-up

Drop a video here or click to choose

MP4, MOV, WebM, MKV, AVI ยท any file size

Processed in your browser. Files are not uploaded.

Files are processed on your device, in your browser, and are never uploaded. Processing speed and the largest file you can handle depend on your device.

What does Video Compressor do?

Compress video fast in your browser with hardware acceleration. Pick High, Basic or Low compression or a target size, downscale, and compare sizes.

Price
Free
Account
Not required
Processing
Entirely in your browser; your data and files are not uploaded
Works on
Any modern browser on desktop, tablet or phone
Category
Video and Audio Tools

How to use the video compressor

  1. Drop a video (MP4, MOV, WebM, MKV or AVI).
  2. Select a compression level: High (smallest size), Basic (medium size, better quality) or Low (highest quality). Each card shows an estimated final size. Or choose Target size and pick the size you need.
  3. Optionally lower the resolution to 1080p, 720p or 480p, pick the audio bitrate, or remove the audio. Keep Fast mode on for hardware-accelerated encoding.
  4. Click Compress video and follow the live progress bar. When it finishes, check the preview, compare the sizes and download the MP4.

Worked example

Fitting a 60 second clip into 25 MB

25 MB is 25 x 1,048,576 x 8 = 209,715,200 bits. Over 60 seconds that is 3,495 kbps in total. The tool keeps 3% for container overhead and subtracts 160 kbps of audio, so it asks the encoder for 3,230 kbps of video. The page shows the estimated size before encoding and the real size afterwards.

How it works

Fast mode (default): the video is decoded and re-encoded to H.264 with the browser's WebCodecs API, which uses your device's hardware video encoder where available; Mediabunny handles reading and writing the files. The bitrate for High, Basic and Low is width x height x 30 fps x 0.045, 0.075 or 0.11 bits per pixel, capped at 35%, 50% or 70% of the original video bitrate, and the final size shown on each card is calculated from it. In target size mode the bitrate comes from the formula below, and if the hardware encoder overshoots the target by more than 3%, the video is encoded once more at a proportionally lower bitrate. HDR video, AVI files and anything the browser cannot encode use the standard engine automatically. Standard engine: video is re-encoded with libx264 (H.264, -preset faster by default, yuv420p) into an MP4 with AAC audio. High, Basic and Low use constant rate factor: -crf 28, 23 or 20. Target size mode computes video kbps = target MB x 1,048,576 x 8 / 1000 / duration x 0.97 - audio kbps and encodes in two passes with -b:v (single pass adds a 1.5x -maxrate and 2x -bufsize cap). The frame rate and aspect ratio are kept, phone rotation is applied so the video plays upright, and downscaling uses Lanczos filtering to set the shorter side to 1080, 720 or 480 pixels with even dimensions. Colour tags are kept (untagged HD video is tagged BT.709), and HDR video (HLG or PQ, as recorded by recent iPhones) is tone mapped to standard range. AAC audio at or below the chosen bitrate is copied; other audio is encoded to AAC. -movflags +faststart lets the MP4 start playing before it fully downloads.

Assumptions

  • Final sizes on the level cards are estimates that assume 30 fps; the real size is shown after compressing.
  • The target size is an average-bitrate goal. With the standard engine, two-pass landed at 97% of the target in our tests, while single pass came out at 67% on a clip with a still section, so two-pass is on by default there.
  • 1 MB here means 1,048,576 bytes, as shown by Windows and this site.
  • The audio bitrate you choose is reserved in the size maths. AAC audio already at or below it is copied, and its real bitrate is used instead.
  • The duration comes from the file; if it has none (some screen recordings), the tool reads the whole file once to measure it.

Frequently asked questions

Are my files uploaded?

No. The file is processed by FFmpeg compiled to WebAssembly, running inside your browser. Only the engine itself is downloaded (once, from cdn.jsdelivr.net); your video, audio or images never leave your device.

How fast is it?

With Fast mode, the hardware video encoder does the work. In our test a 20 second 1080p clip (36.6 MB) compressed to 12.5 MB in 12 seconds, while the standard x264 engine was on course to take about 6 minutes for the same clip. Speed depends on your device and browser.

Why is my compressed video bigger than the original?

If the original is already efficiently compressed at a low bitrate, re-encoding can produce a larger file. Try High compression, a lower resolution, or a target size below the original size.

What is CRF?

Constant rate factor is the x264 quality setting used when Fast mode is off. Lower numbers mean higher quality and bigger files; each step of about 6 roughly halves or doubles the bitrate. The final size depends on how complex the video is, so it cannot be predicted in this mode.

Which encoding speed should I use with Fast mode off?

Balanced (x264 faster) is the default. In our tests it made files about 10% smaller than Fast (veryfast) at the same measured quality, taking about twice as long. Smallest file (medium) saved about 3% more than Balanced and took about 1.4 times as long.

Why does the first run take longer?

The first time you use any of the video and audio tools, your browser downloads the FFmpeg engine, about 31 MB. It is stored in your browser, so later runs on this device start without downloading it again unless you clear site data.

Limitations

  • FFmpeg runs as single-threaded WebAssembly in the browser, which is much slower than desktop FFmpeg. Processing time depends on the file length, resolution and your device; long HD videos can take many minutes.
  • There is no file size limit. Browsers do limit how much memory a page can use. Very long or high resolution files can fail with an out of memory message, especially on phones and tablets.
  • Output is always MP4 (H.264 video, AAC audio). Subtitles and extra audio tracks are not kept.
  • Very small targets for long videos are rejected because the picture would be unwatchable; the tool shows the minimum size for your clip.