You record something on your phone, try to send it, and are told it is too large. The photo you took ten seconds earlier went through without complaint and was two hundred times smaller.
Video is not a heavier kind of image. It is thirty of them a second, and the size follows from that in a way that is genuinely simple once somebody writes it down.
The tool Compress Video Shrink a clip until it fits WhatsApp, email or Discord. Pick the destination, not a bitrate.The four numbers
Every video file’s size comes from these, and nothing else:
- Resolution — how many pixels in each frame. 4K is four times 1080p, not twice.
- Frame rate — how many of those frames per second. 60 is double 30.
- Bitrate — how much data per second the encoder is allowed to spend.
- Length — how many seconds of it there are.
The useful part is that the last two are the whole answer. Bitrate × length = file size, near enough, and the arithmetic is one line: 8 megabits per second for 60 seconds is 480 megabits, which is 60 megabytes.
Resolution and frame rate matter because they decide what bitrate the video needs. More pixels, more frames, more data per second to describe them at the same quality.
What a phone actually records
You do not have to take anybody’s word for these — your phone will tell you. On an iPhone they are printed in Settings → Camera → Record Video, and Android phones show something similar.
| Setting | Roughly, per minute |
|---|---|
| 720p at 30 fps | 40 MB |
| 1080p at 30 fps | 65 MB |
| 1080p at 60 fps | 100 MB |
| 4K at 30 fps | 170 MB |
| 4K at 60 fps | 400 MB |
A three-minute clip at the top setting is over a gigabyte. Nothing has gone wrong; that is simply what the setting means, and most phones ship with a high one because it demonstrates the camera well in a shop.
Which is worth knowing before you record rather than after: if the video is for a message or a form, changing that setting once solves the problem permanently.
Container and codec: the misunderstanding that wastes the most time
This is the one nobody explains, and it causes more confusion than everything else on this page together.
MP4, MOV, MKV, AVI and WebM are containers. Boxes. They hold a video stream, one or more audio streams, subtitles and timing information, and they say almost nothing about how the video inside was compressed.
H.264, H.265, VP9 and AV1 are codecs. They are the compression itself — the thing that decides both the quality and the size.
Two consequences follow, and both catch people out:
- Renaming a file changes nothing. Calling a .mkv a .mp4 relabels the box. The contents are identical, and if the player refused it before, it will refuse it now.
- “Converting” is sometimes just repacking. Some converters move the same streams into a different container without touching them. That is fast, lossless and occasionally exactly right — and it also means the file is not smaller, and can be unplayable if the new container is not allowed to hold that codec. A file that looks perfectly valid and plays nowhere is the usual result.
If a conversion finished in two seconds, nothing was re-encoded.
So what does compressing actually do?
It decodes every frame and encodes them all again at a lower bitrate — often at a smaller resolution too. There is no way around that. A video cannot be shrunk the way a zip file compresses a document, because it is already compressed to within an inch of its life.
That is why it is slow, and why the time depends on the length of the video rather than on how big the file is. Doing it in a browser is slower still — the encoder is running as WebAssembly on one thread rather than as native code on all of them. That is the price of the file never leaving your machine, and for a clip of a few minutes it is a fair one.
The order that works
1. Cut it first. Size is proportional to length, so removing the twelve seconds of walking towards the subject removes twelve seconds’ worth of megabytes, instantly and losslessly. Nobody does this first and it is the cheapest saving available.
2. Drop the resolution. 4K down to 1080p is a quarter of the pixels. On a phone screen — where the video will almost certainly be watched — the difference is close to invisible.
3. Halve the frame rate. 60 fps to 30 is half the frames. Keep 60 for sport, gameplay and anything in slow motion. For a person talking, a view, a room, a document, 30 is what television used for decades.
4. Then lower the bitrate, which is what a compressor’s quality setting really controls. Do this last, because by now the video needs far less of it.
5. Consider whether you need the picture or the sound at all. If only the audio matters — a lecture, an interview, a song — taking the audio out turns hundreds of megabytes into a few. If only the picture matters, removing the audio track saves less but is instant, and it is often the point rather than the size.
H.264, H.265 and the licensing story again
H.264 is the codec almost everything records and almost everything plays. It is twenty years old and it is the safe choice.
H.265 — also called HEVC — produces roughly the same quality in about half the space. It is also patented, which is why support for it is patchy in exactly the places you would want to send a file: some browsers, some editors, some upload forms. It is the same licensing story as HEIC photographs, for the same reason: HEIC is HEVC with a single frame in it.
AV1 is the newer, royalty-free answer, better again and supported by every current browser, though thinner outside them.
The practical rule has not changed in a decade: H.264 for a file you are sending to somebody, anything newer for a file you control both ends of.
And a GIF is not a small format
Worth saying because it reliably surprises people: a GIF of a five-second clip is frequently larger than the video it was made from. GIF is over thirty years old, limited to 256 colours, and has none of the tricks a video codec relies on — it cannot say “most of this frame is the same as the last one” in any efficient way.
Making one is a decision about where it can be posted and whether it should play by itself, not a way of saving space. Keep them short, keep them small, and expect the file to be bigger than instinct suggests.
Targets worth aiming at
Gmail refuses attachments over 25 MB and offers a Drive link instead, which is a published limit you can plan around. Most chat apps are far stricter and do not publish a figure at all, so aiming under about 15 MB for anything sent through one is a reasonable, if unofficial, target.
For a clip going onto a website, resolution matters more than the number: 1080p at a moderate bitrate looks right on every screen anybody will use, and 4K on a web page is bandwidth spent on something almost nobody will see.
One thing not to do
Do not compress the compressed copy. Every re-encode works on the previous encode’s damage as well as the picture, and video shows this faster than photographs do — blocky skies, smeared movement, edges that swim.
If an attempt came out too large, go back to the original and compress that harder. Keep the original until you are certain, because nothing recovers what a re-encode threw away.