Why zipping a video does almost nothing

Zipping a video saves about 1%. Sometimes it makes the file slightly larger. This surprises everyone the first time, because zipping a folder of documents routinely cuts it by 70%, and a video is just another file.

It is not just another file. It is already compressed, by something far more aggressive than ZIP, and that is the whole story.

What ZIP actually does

ZIP, 7z, RAR and the rest are lossless, general-purpose compressors. They work by finding repetition: a sequence of bytes that appeared before gets replaced with a short reference back to the earlier copy. Text is full of repetition (the same words, the same spaces, the same letter frequencies) so a text file shrinks enormously. A spreadsheet, a log file, a program's source code: all the same.

The method is lossless, which means the output decompresses to exactly the input. That is a requirement for documents and a guarantee that nothing useful can be thrown away.

What a video file already is

A video from a phone, a camera, a screen recorder or a download is H.264, H.265 or AV1 inside an MP4 or MOV container. Those codecs have already done something far beyond what ZIP can:

• Every frame has been transformed and quantised so that detail the eye cannot see has been discarded. That is lossy, and it is where the big savings are.
• Each frame has been described as a difference from the frames around it, so a static background costs almost nothing after the first time.
• What remains has been entropy-coded, which is the same trick ZIP uses, applied already.

The result is a stream of bytes with almost no repetition left in it. By design it looks like random noise to anything that comes after, because every pattern that could be exploited already was. ZIP goes looking for repetition, finds none, and gives up. The 1% it saves is the container's metadata and index, which were stored as plain structured data.

The same is true of JPEG photographs, MP3 and AAC audio, and most PDFs. If a format is already lossy-compressed, a second lossless pass has nothing to work with.

The numbers

File	Original	Zipped	Saving
10-page text document (.txt)	48KB	15KB	69%
Word document (.docx, already a ZIP)	320KB	290KB	9%
Folder of CSV exports	12MB	1.8MB	85%
1080p phone video (.mp4)	184MB	183MB	0.5%
JPEG photo	4.2MB	4.1MB	2%
Scanned PDF (JPEG pages inside)	9MB	8.9MB	1%

A .docx is in that table on purpose: it is already a ZIP archive with a different extension, which is why zipping one again barely moves.

When zipping a video is still worth doing

Three cases, none of them about size.

The destination refuses .mp4 but accepts .zip. Some ticketing systems, forms and corporate mail filters allow archives and block media. Zipping is a wrapper, not a compressor, but a wrapper is what is being asked for.

Several files need to travel as one. A dozen clips as a single attachment is more convenient to send and to receive, and the archive costs nothing.

The file is not actually a video. Raw uncompressed capture (some screen recorders and camera formats produce it, and anything described as "uncompressed AVI" certainly is) zips well, because nothing has compressed it yet. If you have a 2GB file for 30 seconds of footage, that is what you have, and the right fix is still to encode it to H.264 rather than zip it.

What to do instead

Re-encode. The file is compressed already; the question is whether it is compressed as hard as the situation needs.

A phone encodes in real time with no time to search for the best description of each frame, so it spends two to four times the bits a careful encode would. Why a phone video is as big as it is comes down to four numbers, and bitrate is the one worth changing first.

The arithmetic for a target size:

bitrate available = (limit in bytes x 8) / duration in seconds

184MB video, 2 min 10 s, needed under 25MB:
  25,000,000 x 8 / 130 = 1.54 Mbps

That is tight for 1080p and fine for 720p. The encode takes seconds on hardware and the result is a quarter of the size, at a quality the recipient will not distinguish on a phone. ZIP would have delivered 183MB.

One caution. An email limit measured against the attachment is not what you think: Gmail's 25MB is measured after base64 encoding, so the real ceiling is about 18MB, and GitHub states its 25MB and 100MB limits in MiB, which is 4.9% less than the number suggests. Zipping does not help with either, and a re-encode that targets the published figure rather than the real one still fails.

Compress a video to 25MB, to 18MB or to 100MB does the division above, encodes on your own machine, measures the real output and re-encodes if it came in over. The video is never uploaded.

The one-line version

ZIP finds repetition. A video has had its repetition removed by a codec already, so ZIP finds nothing. To make a video smaller you have to run a codec again, with less bitrate, and that is a different operation with a different name.

FAQ

Does zipping a video reduce its file size?

By about 1%, and occasionally it gets larger. Video is already compressed by a codec (H.264, H.265 or AV1) that removes every pattern ZIP could exploit, so a general-purpose compressor has nothing left to find. The same is true of JPEG photos, MP3 audio and most PDFs. To shrink a video you need to re-encode it at a lower bitrate, not archive it.

Why does zipping work on documents but not on videos?

Because documents contain repetition and compressed video does not. ZIP replaces repeated byte sequences with references to earlier copies. Text, spreadsheets and logs are full of repeats. A video codec has already found and removed every repeat across frames and within them, and what remains is close to random, which is incompressible by design.

Can I zip a video to email it?

You can, but it will not fit any better than the original. A 60MB video is a 59MB zip, still far over Gmail's effective 18MB ceiling. Zipping only helps when the mail system blocks .mp4 attachments and allows .zip, which is a filter problem rather than a size problem. For size, re-encode the video to a target under the limit.

What is the difference between compressing a video and zipping it?

Zipping is lossless and general-purpose: the output unpacks to exactly the input, and it only works where there is repetition to remove. Compressing a video means re-encoding it with a codec at a lower bitrate, which is lossy: it discards detail the eye cannot see and produces a smaller video that plays directly. The second is what people mean when they need a video to fit a limit.