"Without losing quality" is doing a lot of work in that sentence, so let us be straight about it up front: every practical video compressor is lossy. Information is discarded. The honest question is not whether anything is lost — it is whether you can see what was lost, and that depends far more on your source file than on the tool you use.
The good news is that most videos carry an enormous amount of slack. Phone recordings, screen captures and exports from editing software are routinely encoded well above what their content needs. Removing 70% of the file changes nothing you can perceive. Footage that is already tightly compressed has no slack left, and pushing it will show.
This guide covers how to tell which one you have, and what to do about it.
Quick takeaways
- Four things control file size, in rough order of impact: resolution, bitrate, duration, codec.
- Dropping the resolution is the single biggest win and usually the least visible one, if the video will be watched on a phone.
- Screen recordings compress dramatically better than camera footage. In our tests, a screen capture held VMAF 89.9 at 0.79 MB where camera footage fell to 74.4 at three times the size.
- Trim before you compress. Thirty seconds you cut is thirty seconds of bits you never have to spend.
- Stay on H.264 if the file is going to someone else. H.265 is ~30% smaller and will not play on plenty of devices.
- If you have a hard limit to clear, use a target size mode rather than guessing at quality settings.
First: why is your video big?
Four factors, and they are not equally important.
1. Resolution — the biggest lever
A 4K frame carries four times the pixels of a 1080p frame. A 1080p frame carries just over twice a 720p one.
Here is the thing people miss: resolution only matters relative to where the video will be watched. A 4K clip that will be viewed in a phone-sized window in a chat app is carrying millions of pixels that the screen will throw away anyway. Scaling it to 1080p before sending costs you nothing visible and is the single largest saving available.
Rule of thumb:
- Going to someone's phone, in a messaging app → 720p is plenty
- Going on a website as an inline video → 1080p
- Going to a TV, or being edited further → keep the original
2. Bitrate — where the slack lives
Bitrate is how many bits per second the encoder spends. It is where "badly encoded" shows up, and it is why so many files compress so well.
Your phone does not carefully optimise. It encodes fast, in real time, with a generous bitrate, because it has a battery to preserve and no idea what the footage contains. Screen recorders are worse. Export presets in editing software are frequently set to "high quality" and left there.
All of that is slack. A re-encode that takes thirty seconds instead of real-time can describe the same content in far fewer bits.
3. Duration — the one people forget
Bits per second times seconds. Cutting the dead thirty seconds at the start of a screen recording removes those bits entirely, with no quality cost at all, because they were never encoded.
Trimming is the only "compression" that is genuinely free. Do it first.
4. Codec — real but oversold
H.265 (HEVC) produces files roughly 30% smaller than H.264 at matching visual quality. That is a genuine improvement and it matters when you are archiving a lot of footage.
It also encodes about 3.8x slower, and older phones, older TVs and some browsers will simply refuse to open it. If the file is going to someone else's device, stay on H.264. The 30% is not worth a message saying "it won't play for me".
What about frame rate?
Less than people expect. Halving 60fps to 30fps does not halve the file — the encoder was already describing the second frame of each pair as "mostly like the last one". You will save far more by dropping one resolution step, and you will not make motion look choppy doing it.
How much can you actually save?
This depends on your source far more than on your settings, and we have measured how much.
We encoded three 1080p clips at six quality levels each and scored every result with VMAF, Netflix's perceptual quality metric. Around 93 is generally treated as the point where most viewers stop noticing a difference.
| Source type | Best quality setting | A hard compression setting |
|---|---|---|
| Screen recording | 2.09 MB · VMAF 96.7 | 0.79 MB · VMAF 89.9 |
| Animation | 15.52 MB · VMAF 96.0 | 1.96 MB · VMAF 79.7 |
| Camera footage | 12.74 MB · VMAF 97.2 | 2.43 MB · VMAF 74.4 |
All three clips are ten seconds and 1080p, so those numbers are directly comparable.
Read the two bold rows together. At the same aggressive setting the screen recording is three times smaller than the camera footage and scores 15.5 VMAF points higher. Screen content is mostly flat colour and mostly unchanged between frames — an encoder can say "identical to last frame" in almost no bits. A moving scene with sensor noise has genuinely new information in every single frame.
So when a compressor promises "up to 90% smaller", that number is real, and it is real for screen recordings. It is not real for your handheld holiday footage. Anyone quoting one number for both is not telling you the whole thing.
The full measurements, method and limitations are in CRF explained.
The practical method
Step 1 — trim anything you do not need
Free. Do it before anything else.
Step 2 — decide where this video is going
This determines everything downstream, and it is the step people skip.
- A hard upload limit? Skip to the target-size approach below.
- Watched on a phone? Scale to 720p and stop worrying about the rest.
- Going on your website? 1080p, and get it well under 10 MB — a heavy hero video is one of the most common causes of a poor Core Web Vitals score.
- Archiving? Keep 1080p, use a stronger quality setting, and consider H.265 since only you will play it back.
Step 3 — pick resolution before quality
If the destination does not need the pixels, drop them first. Scaling 4K to 1080p typically removes more than half the file before you have touched a quality slider, and on a phone screen the result is indistinguishable.
Step 4 — then adjust quality
Now use a preset or a CRF value. Start moderate. Look at the result. Adjust once. Two minutes of actually looking beats any recommended number, including the ones in this article.
Step 5 — check the difficult parts, not the average
If you are going to spot-check, check the hardest few seconds: fast motion, rain, confetti, smoke, water, fine texture. Those are where compression artifacts show up first. A still shot of someone talking will look fine at almost any setting and tells you nothing.
When you have a hard limit
If the requirement is "this must be under 20 MB", quality settings are the wrong tool — you would be guessing, encoding, measuring, and guessing again.
Use a target-size mode. It encodes in two passes: the first measures how demanding the content is, the second spends exactly the budget available. It lands within a couple of percent of the number you typed.
One caution: aim slightly under the ceiling. Landing 2% over a hard limit means the upload fails, and some platforms count bytes differently than your file manager does.
The common limits, each with the arithmetic worked out for how long a clip fits:
- Discord — 20 MB free (raised from 10 MB in August 2026), 50 MB Nitro Basic, 500 MB Nitro
- Email — Gmail advertises 25 MB but measures the base64-encoded attachment, so the real ceiling is about 18 MB
- WhatsApp — 16 MB in chat, or 2 GB sent as a document with no re-encoding
- TikTok — 72 MB from the Android app, 288 MB from iOS
- Instagram — 4 GB, but they re-encode everything, so what you hand them matters more than the cap
- Telegram — 2 GB free, 4 GB Premium
- Slack — 1 GB on every plan
- X (Twitter) — 512 MB, but 2:20 of runtime binds first
Mistakes that cost you quality for nothing
Compressing twice. Every lossy pass discards information permanently. If you compress, look at it, and compress the result again, you are stacking losses. Go back to the original and use a stronger setting once.
Compressing before editing. Edit the original, export once at the end.
Chasing a smaller number than you need. If the limit is 20 MB, do not target 8. There is no prize.
Uploading a huge master to a platform that re-encodes. Instagram, TikTok and WhatsApp all re-encode whatever you give them. Handing their encoder a 3 GB file does not preserve quality — it gives their encoder more to throw away, and it throws away whatever it likes. A clean, correctly sized file survives better.
Using H.265 for something you are sending to someone else. Covered above, and it remains the most common cause of "it won't open".
FAQ
Can you really compress a video without losing quality?
Not literally — lossy compression discards information by definition. But whether you can see the loss is a different question, and for most source material the answer is no. Phone recordings and screen captures carry so much slack that removing 70% of the file changes nothing perceptible. Footage that was already compressed hard has no slack, and pushing it will show.
What is the best format for a compressed video?
MP4 with H.264 video and AAC audio. It plays on every phone, browser, TV and editing program in current use. MKV and WebM have real advantages in specific situations, but if the file is leaving your machine, MP4 is the answer.
Should I lower the resolution or the bitrate?
Resolution first, if the destination does not need the pixels — it is the larger saving and usually the less visible one. Once the resolution matches the destination, tune the bitrate or quality setting.
Does compressing a video reduce its length or resolution?
No. Compression re-encodes the same frames more efficiently. Length is unchanged, and resolution only changes if you explicitly ask for it. Those are separate operations that often live in the same tool.
Why is my compressed video still too big?
Usually one of three things: the source is already efficiently encoded and has no slack; the clip is simply long, and duration multiplies everything; or the resolution is higher than the destination needs. Check duration and resolution before reaching for a more aggressive quality setting.
Is it safe to compress video online?
It depends entirely on the operator, so ask them what they actually do with your file. Ours: your browser uploads over a one-time signed link to storage we control, the encoder pulls it server-to-server, and the staged copy is deleted the moment the job settles. Results expire after two hours for anonymous jobs, up to seven days when signed in. We do not watch, train on, share or resell anything.
Compress a video now — no account, no watermark, nothing to install. Or pick your upload limit and we will set the target for you.