How to make a video file smaller without ruining it

Video is the only everyday file type where "a bit smaller" is not a setting but a trade. Every megabyte you remove comes out of length, resolution, bitrate or codec efficiency, and the right choice depends on where the file is going. A clip for an email attachment, a video for a website and a recording you want to archive need different answers. This guide gives you the four levers, the arithmetic for a hard limit, and the order to pull them in.

The four levers

Length. Half the duration, half the file, with no loss of quality at all. If the recipient needs forty seconds out of six minutes, cutting is worth more than any encoder setting. Cutting with the trimmer on this site copies the selected bytes out without re-encoding, so it takes about a second and keeps the picture exactly as it was.

Resolution. 1080p has 2.25 times the pixels of 720p and four times those of 540p. Fewer pixels need fewer bits to look clean, and the screen the video will be watched on is often smaller than the recording anyway: a chat window, a phone held upright, a slide. Dropping 4K phone footage to 1080p is usually invisible to the viewer and cuts the file by two-thirds or more.

Bitrate. This is the number of bits the encoder may spend per second of video, and it is the size dial. Lower it and the encoder starts throwing away fine detail: grass becomes a green smear, text edges shimmer, dark scenes go blocky. The art is lowering it only as far as the content allows, and that depends on how much motion there is. A talking head at 720p looks fine at 1.5 Mbps; a football match at the same bitrate looks bad.

Codec. H.264 is the format that plays everywhere: every phone, browser, TV and messaging app. Newer codecs such as HEVC and AV1 produce smaller files for the same quality but do not play everywhere, and a file that does not open is the largest possible failure. The compressors here always output H.264 in an MP4 container for that reason, and converting an iPhone HEVC recording to it is often a size reduction in itself because the encoder gets to choose a sensible bitrate.

The arithmetic for a hard limit

File size is bitrate multiplied by duration, so a limit in megabytes and a clip length give you the bitrate you have to hit:

bitrate (Mbps) = size (MB) × 8 ÷ seconds

A 25 MB email limit and a two-minute clip: 25 × 8 ÷ 120 = 1.67 Mbps in total. Reserve about 0.13 Mbps for the audio track and the video gets 1.5 Mbps, which is comfortable at 720p. The same limit with a ten-minute clip gives 0.33 Mbps: that only works at 480p or lower, and the honest advice is to trim or send a link instead.

Where it is going	Limit (at time of writing)	Realistic length at good quality
Gmail attachment	25 MB	about 2 to 3 minutes at 720p
Outlook.com attachment	20 MB	about 2 minutes at 720p
Discord, free account	20 MB (since August 2026)	about 2 minutes at 720p
WhatsApp, sent as video	re-compressed by the app; 16 MB was the long-standing cap	about 1.5 minutes at 720p
Website background video	no cap, but every visitor downloads it	keep it under 5 MB, muted, 10 to 20 seconds, looped

The Compress Video to Size tool does this calculation for you: type the megabytes, and it reads the duration, works out the bitrate, budgets for audio and the container, and encodes to that number, lowering the resolution when the bitrate gets too thin for the pixel count. If the first pass lands over, it measures by how much and re-encodes once with a corrected rate. If the file is already under the target it hands it back untouched.

The order to do things in

1. Trim. Remove the dead start and end and anything the recipient does not need. Free, instant, lossless.
2. Compress to the number if there is a limit, or to a quality level if there is not. The exact-size tool picks the resolution for you from the bitrate it has to hit (720p or 480p when the budget is thin, the original size otherwise). The Compress Video tool keeps the original resolution and offers a few quality presets from "small" to "high quality"; Balanced is the sensible first try.
3. Check it plays. Open the result before sending it. MP4 with H.264 and AAC audio plays on anything made in the last fifteen years.

When the picture does not matter

A recorded lecture, a meeting, an interview: often the sound is the content and the picture is a static room. Extracting the audio with Video to MP3 turns an hour-long 1.5 GB recording into a 60 MB file that carries everything of value, and that file is small enough to attach, archive and transcribe.

What "no quality loss" really means

Any re-encoding of a lossy video loses something; a compressor that promises otherwise is describing the result as it looks, not as it is. The useful distinction is between visible and measurable loss. Dropping a 50 Mbps phone recording to 8 Mbps at the same resolution is measurable and invisible. Dropping it to 0.5 Mbps is visible to anyone. The exception is trimming by stream copy, which really is lossless, because the bytes are copied rather than decoded and re-encoded, at the price of the cut landing on the nearest keyframe rather than the exact frame.

Why none of this needs an upload

Encoding video on a server means uploading the full original first, which for a 1.5 GB recording is the slowest step of the whole process and leaves a copy of your footage on a machine you do not control. Every video tool on this site runs FFmpeg inside your browser: the engine downloads once (about 31 MB) and then the work happens on your own processor. Your file never leaves your device, which is why there is no size limit, no watermark and no queue. The cost is that your hardware sets the speed: a laptop encodes a few minutes of 1080p in a couple of minutes, a phone takes longer, and the tab needs to stay open while it works.