Video Bitrate
Video bitrate is the data budget for the encoded picture stream and can be a target, a limit, or a measured result.
| Average video rate | Bytes for one second of pictures | Video-only size over two hours |
|---|---|---|
| 4 Mb/s | 0.5 MB | 3.6 GB |
| 8 Mb/s | 1 MB | 7.2 GB |
| 16 Mb/s | 2 MB | 14.4 GB |
Decimal units. Two hours is 7,200 seconds; the calculation excludes soundtracks and packaging.
Video bitrate counts coded picture data against media duration. It is the video's data budget or measured rate, rather than the whole file's rate when that file also contains audio and other tracks.
Divide one second among its pictures
At 8 Mb/s, one media second contains 1,000,000 video bytes on average.
At 24 fps, the average allocation is 1,000,000/24 =41,666.67 bytes per picture. At 60 fps it is 16,666.67 bytes. The same per-second budget is spread among more pictures.
Actual pictures need not have those average sizes. Consider a deliberately constructed 24-picture interval:
| Pictures | Bytes per picture | Combined bytes |
|---|---|---|
| One independently coded picture | 195,000 | 195,000 |
| 23 predicted pictures | 35,000 | 805,000 |
| Total | 1,000,000 |
The interval still has 8 Mb/s of coded video. The large picture establishes content that the following pictures can reuse. Their sample count is not smaller; their encoded descriptions are.
This example explains why a source can contain substantial frame-size variation at a stated average bitrate.
Decide what the encoder is trying to hold
A target average rate tells rate control to allocate data around an overall budget. It may use more in difficult sections and less elsewhere.
A rate/buffer limit constrains local coded demand. Its meaning includes the relevant time and buffering model, not only a maximum number printed beside a file.
A quality-based encode lets the resulting average rate emerge from its source and target. The measured bitrate is then an outcome rather than an exact amount selected in advance.
CBR, VBR, and CRF develop these policies separately.
Connect the budget to the source
An encoder can reuse a wall across neighboring pictures with small corrections. Grain, fine texture and changing scenery require more information. Adding more frames also creates additional prediction opportunities, so a simple bits-per-picture division does not by itself determine visible quality.
Presets change how thoroughly the encoder searches for a representation; quantization changes what approximation it accepts. The rate budget constrains their result.
In an Emby session, compare the source video's measured rate with the delivered video's rate. Video conversion can create a different allocation while the original library file keeps its own rate. A file-wide bitrate can also include unused soundtracks, so keep the measurement's counted streams and duration explicit.