Why videos are so large, and what compression does
A minute of 1080p footage from a phone is commonly 100MB or more. Cameras encode generously on purpose — the recorder has to keep up in real time on a battery, so it spends bitrate rather than CPU cycles. Re-encoding afterwards, with no real-time constraint, is where the savings are.
Compression works by discarding information: fine detail in areas the eye tends not to scrutinise, and differences between frames that are nearly identical. Done well it is close to invisible. Done hard it shows up as blockiness in shadows and smearing on fast motion.
Why the setting is quality, not size
The control here is CRF, the same knob professional encoders expose. It holds picture quality constant and lets the bitrate vary with how demanding each scene is.
This matters because footage varies enormously. A screen recording of a mostly-static document has almost nothing changing between frames and compresses to a fraction of its size. Handheld footage of moving leaves has detail changing in every pixel of every frame and compresses barely at all. A target-size mode would have to pick one bitrate and apply it to both, wasting it on the first and starving the second.
- High — near-transparent quality. Modest savings, safe for footage you will edit later.
- Balanced — the usual choice. Large savings, differences you have to look for.
- Small — visible softening on detailed footage. For when a hard size limit has to be met.
Being honest about the speed
This is genuinely slow, and it is worth saying why rather than hiding it. Compressing means decoding every frame and encoding it again. Native FFmpeg does that fast partly by using your CPU's vector instructions and often a dedicated hardware encoder. A WebAssembly build inside a browser tab can reach neither, and runs roughly an order of magnitude slower as a result.
The trade is straightforward: a cloud service is much faster because it is a rack of servers, and the price is that your video is on those servers. Here you pay in time and keep the file.
One practical note: keep the tab visible. Browsers throttle background tabs aggressively, and a hidden tab can slow to a crawl or stall.
When compression will not help
If a video has already been through this — downloaded from a streaming site, exported for the web, received through a messaging app — most of the redundancy is already gone. Re-encoding then costs quality without buying much size, and can even produce a larger file if the quality setting targets a higher bitrate than the source used.
The tool detects that case and tells you when the output came out bigger, rather than handing you a worse file and calling it a success.
Common uses
- Getting a clip under an email or messaging attachment limit
- Shrinking phone footage before uploading it over a slow connection
- Reducing a screen recording for a bug report or a support ticket
- Freeing space without deleting recordings outright
- Preparing video for a platform with a strict file size cap