Video Compression Calculator
Work out what a file size leaves you, or what a resolution costs. The bitrate is exact arithmetic; the resolution verdict is a recommendation, and this page keeps the two apart — then shows what we actually measured at that bitrate on three kinds of 1080p footage.
25 MB over 1 min leaves 3,297 kbps for video. The best resolution that clears its comfortable threshold is 720p. 1080p is the highest that is still watchable.
What 25 MB clears
- WhatsApp Video in chat (mobile) 16 MB
- Gmail in practice 18 MB· base64 adds ~37%
- Discord Free 20 MB
- Gmail 25 MB
- TikTok Android app 72 MB
- X (Twitter) Standard 512 MB
What we measured at this bitrate
VMAF from our own encodes at 1080p, libx264 preset medium, measured 2026-09-06. One clip per content type, so treat these as positions rather than tolerances — the gap between them is the finding, not the individual numbers.
- Screen recording≥ 96.74VMAFabove every bitrate we measured for this content (top was 1,750 kbps → 96.74); not extrapolated
- Animation89.1VMAFbetween measured 2,840 kbps → 87.85 and 3,793 kbps → 90.51
- Camera footage85VMAFbetween measured 3,161 kbps → 84.36 and 3,991 kbps → 88.15
The arithmetic
Given a size and a duration there is exactly one answer, and it is a hard ceiling rather than an opinion. Everything the calculator labels “exact” comes from these two lines.
- 8,388,608 is bits per mebibyte. Upload limits are enforced in mebibytes while being called megabytes, so using 1,000,000 would compute a target about 5% larger than the one actually being applied.
- 0.98 reserves container and muxing overhead. MP4 costs roughly 1–2%; this budgets the pessimistic end, because the failure mode worth avoiding is landing just over a hard cap.
- 128 kbps is the default audio assumption — stereo AAC at a sane quality. Speech starts to sound thin below about 96 kbps.
Where the resolution thresholds come from
These are the numbers the calculator compares your budget against. They are recommendations, not measurements: derived from published H.264 bitrate ladders — YouTube’s recommended upload bitrates and Apple’s HLS authoring specification — and pulled down by roughly a third. Those ladders describe a first encode of a clean master, while compressing an existing MP4 starts from footage that has already been through one lossy pass and survives a lower bitrate than a master does. Deliberately conservative: the failure we care about is telling someone their video will look fine when it will not.
| Resolution | Comfortable | Watchable |
|---|---|---|
| 4K | 20,000 kbps | 12,000 kbps |
| 1440p | 9,000 kbps | 5,500 kbps |
| 1080p | 4,000 kbps | 2,000 kbps |
| 720p | 2,200 kbps | 1,100 kbps |
| 480p | 1,000 kbps | 500 kbps |
| 360p | 600 kbps | 300 kbps |
At 30 fps. A 60 fps clip of the same resolution wants more, and lowering frame rate is usually the wrong lever — the cost lands on motion, where viewers notice most.
What we measured, and why one bitrate cannot serve everyone
Eighteen encodes: three kinds of 1080p footage at six quality settings each, libx264 preset medium, VMAF scored against the original every time. Measured 2026-09-06. The reason it is on this page is the spread — the three content types sit in completely different places at the same bitrate, so a calculator that returns one verdict is telling at most one of them the truth.
Screen recording
measured 664–1,750 kbps
- 1,750 kbps96.74
- 1,507 kbps96.19
- 1,244 kbps95.49
- 1,012 kbps94.16
- 887 kbps93.23
- 664 kbps89.88
Animation
measured 1,646–13,021 kbps
- 13,021 kbps96.02
- 9,550 kbps95.1
- 5,970 kbps93.3
- 3,793 kbps90.51
- 2,840 kbps87.85
- 1,646 kbps79.67
Camera footage
measured 2,036–10,673 kbps
- 10,673 kbps97.18
- 8,261 kbps95.6
- 5,705 kbps92.48
- 3,991 kbps88.15
- 3,161 kbps84.36
- 2,036 kbps74.38
What this does not cover, so nobody builds a claim on it: one clip per content type, ten seconds each, all 1080p and all H.264. Enough to show that the three classes sit in completely different places; not enough to quote a tolerance on any single number, and it says nothing about H.265. The calculator interpolates between these points and refuses to extrapolate past them.
Quick reference
The highest resolution that clears its comfortable threshold, for common sizes against common lengths. Assumes 128 kbps of audio. This is the calculator’s answer for the combinations people ask for most, worked out in advance.
| Size | 15s | 30s | 1 min | 2 min | 5 min | 10 min |
|---|---|---|---|---|---|---|
| 8 MB | 1080p | 480p | 360p | — | — | — |
| 10 MB | 1080p | 720p | 480p | — | — | — |
| 16 MB | 1080p | 1080p | 480p | 360p | — | — |
| 25 MB | 1440p | 1080p | 720p | 480p | — | — |
| 50 MB | 4K | 1440p | 1080p | 720p | 480p | — |
| 100 MB | 4K | 4K | 1440p | 1080p | 720p | 480p |
| 200 MB | 4K | 4K | 4K | 1440p | 1080p | 720p |
A dash means no resolution on the ladder clears its comfortable threshold at that combination — the clip is too long for the size. Trim it, or accept a soft result.
Calculator questions
How do I calculate the bitrate for a target file size?
Multiply the size in mebibytes by 8,388,608 to get bits, divide by the duration in seconds, then divide by 1,000 for kbps. Subtract the audio bitrate to get what is left for video, and take off about 2% for container overhead. That is the whole calculation, and it has exactly one answer — the judgement only starts when you ask whether that bitrate is enough.
Is the resolution recommendation measured or estimated?
Estimated, and labelled as such throughout. The thresholds come from published H.264 bitrate ladders — YouTube’s recommended upload bitrates and Apple’s HLS authoring specification — pulled down by roughly a third, because those ladders describe a first encode of a clean master and compressing an existing MP4 starts from footage that has already been through one lossy pass. The bitrate figures on this page are arithmetic; the pass/fail verdicts are a recommendation.
Why does the same bitrate look fine on one video and bad on another?
Because content decides how far a bitrate goes. We encoded three kinds of 1080p footage at six quality settings each and scored VMAF on all eighteen: at around 3,300 kbps a screen recording sits above every bitrate we tested for it, animation lands near 89, and handheld camera footage near 85. At the bottom of the range the gap widens to more than fifteen points. A single recommended bitrate cannot serve all three, which is why this page shows all three.
Should I lower the resolution or the quality setting?
Resolution, almost always, and earlier than feels natural. Because pixel count is two-dimensional, 720p needs about 44% of 1080p’s pixels, so dropping a rung buys far more than squeezing the quality setting — and it buys it without the blocking and smearing that over-compression produces. Match the resolution to where the video will actually be watched, then adjust quality.
Does the audio track matter?
On a short clip, a lot. A 15-second video inside 2 MB has about 1,090 kbps to spend in total; a 128 kbps stereo track is nearly 12% of that before the first frame. On anything over a minute it is close to a rounding error. If a clip has no narration, check whether it is carrying a full stereo track of silence — removing it is free.
Why 1 MB = 1,048,576 bytes and not 1,000,000?
Because that is what upload limits are enforced in. Platforms and operating systems overwhelmingly measure in mebibytes while calling them megabytes, so a calculator using 1,000,000 would report a target about 5% larger than the one being enforced — which is exactly the margin that makes an upload fail at the stated limit.
The measurements behind this
- The full CRF sweep these numbers come fromEighteen encodes with VMAF on each, and an explicit account of what one clip per content type can and cannot support.
- VMAFHow to read the scale, and why a mean hides its worst second.
- Which lever to pull once you have the numberResolution and duration usually buy more than the quality setting does.