Video bitrate calculator
Resolution times frame rate times how much the picture moves, times what the codec needs per pixel. That is the whole model; the interesting part is where each number comes from.
Frame size
1920×1080, 16:9. Pick Custom for anything else.
Codec
The universal baseline. Plays on everything.
Motion
General camera footage: documentaries, vlogs, most film and television.
Delivery bitrate
- Frame
- 1920×1080 @ 30 fps
- Pixels per second
- 62.2 M
- Bits per pixel
- 0.088
- Per minute
- 41.3 MB
- Per hour
- 2.5 GB
- Recommended
- 5.5 Mbps
In an FFmpeg command
-b:v 5500k -maxrate 8250k -bufsize 11000kA capped average: the encoder may spend up to 1.5× the target on a hard second, within a two-second buffer. For a file that is not streamed, prefer a quality target instead; see CRF.
How this is calculated
Every bitrate recommendation table is a statement about bits per
pixel per frame. A 1080p30 stream at 5.5 Mbps is spending
5,500,000 ÷ (1920 × 1080 × 30) ≈ 0.088 bits on each
pixel of each frame. That number falls as frames get bigger, because a
larger frame gives the encoder more redundancy to find: the same
ladder gives 720p about 0.11 and 4K about 0.07. The calculator takes
the H.264 delivery ladder below, turns each rung into a bits-per-pixel
figure, and interpolates between them on a log scale for any frame
size, so a custom 1600×900 frame lands between 720p and 1080p rather
than on either.
| Rung | H.264 delivery | Bits per pixel at 30 fps |
|---|---|---|
| 480p (SD) | 1.5 Mbps | 0.122 |
| 720p (HD) | 3 Mbps | 0.109 |
| 1080p (FHD) | 5.5 Mbps | 0.088 |
| 1440p (QHD) | 9 Mbps | 0.081 |
| 4K (UHD) | 17 Mbps | 0.068 |
Frame rate does not scale the bitrate linearly. Consecutive frames at 60 fps are more alike than at 30, so each one is cheaper to predict, and every published ladder lands on about 1.5× for a doubling: YouTube's upload table goes 8 to 12 Mbps at 1080p, 5 to 7.5 at 720p, 35 to 53 at 4K. The calculator uses that curve, which is an exponent of 0.585 on the frame-rate ratio, so 24 fps costs about 12% less than 30 and 120 fps about 2.25× more.
Motion is the roughest dial. A talking head or a screen recording leaves most of every frame unchanged and needs far fewer bits than the model's middle setting; sport, gaming and handheld footage can need twice as many. The three multipliers (0.7, 1.0, 1.4) are the spread recommendation tables use between those extremes, and the honest statement is that content moves the right answer by a factor of two in either direction. If you can, encode a representative minute at a quality target and read the bitrate the encoder chose; the CRF article explains why that is usually the better workflow for files that are not streamed.
Codec factors are the same ones the bandwidth and storage calculators use: HEVC at 60% of the H.264 bitrate for comparable quality, AV1 at 50%. Those are the middle of the published ranges; the measured comparisons on this site show what one encoder at one preset actually delivers on one clip, which is less at generous bitrates and more at starved ones.
Reading the results
The result is a delivery bitrate: what a player receives. Upload targets at YouTube and similar services run about double, because they re-encode what you send and want headroom. An archive or mezzanine master should be higher still, and is usually better made with a quality target than a bitrate at all. The FFmpeg flags under the result express the bitrate as a capped average (1.5× the target over a two-second buffer), which is the shape streaming services expect; drop the cap for a file that will only ever be downloaded.