File size is bitrate times duration divided by eight, and nothing else. Forty-five minutes at 8,000 kbps is about 2,575 MB regardless of whether it is 1080p or 4K — the resolution describes how many pixels the encoder was asked to represent, and the bitrate describes how much data it was allowed to spend doing so.
That is why two files both labelled 4K can differ by a factor of ten, and why the label alone tells you almost nothing about how the footage will look.
How much bitrate does a resolution want?
Roughly width × height × frame rate × bits per pixel. At 1080p30 with a streaming-quality allowance of 0.1 bits per pixel that is 6.22 Mbps — about 466 MB for ten minutes, or 2.8 GB an hour.
| Format | Streaming-ish | Recording-ish |
|---|---|---|
| 1080p30 | ~6 Mbps | 50–100 Mbps |
| 1080p60 | ~12 Mbps | 100–150 Mbps |
| 4K30 | ~25 Mbps | 100–400 Mbps |
| 4K60 | ~50 Mbps | 200–400 Mbps |
The gap between the two columns is the difference between delivering a finished file and capturing one you intend to edit. Recording bitrates are deliberately generous because every subsequent step — colour grading, stabilisation, re-encoding — is easier with more data to work from.
Codec changes the arithmetic substantially. Those figures assume something H.264-like; HEVC and AV1 deliver comparable quality at roughly half the bitrate, at the cost of much more processing to encode.
Where do the size errors come from?
Two ambiguities, both units. The first is bits and bytes: bitrate is quoted in bits per second and file size in bytes, so the division by eight is the step people skip.
The second is smaller and equally persistent. A megabyte is 1,000,000 bytes to a storage manufacturer and 1,048,576 to some operating systems, a difference of about 4.9 per cent — and it compounds to about 7.4 per cent at the terabyte scale. It is why a "128 GB" card formats to about 119 GB.
Both together explain most of the gap between a calculated recording time and the real one. A 128 GB card at 100 Mbps calculates to about 2 hours 51 minutes and delivers rather less.
How do you hit a size target?
Divide the target by the duration and encode in two passes. Fitting ten minutes into 500 MB needs about 6,990 kbps total including audio — so around 6,700 kbps of video once a 256 kbps audio track is accounted for.
A single pass has to guess how to allocate bits before it has seen the footage, and it typically overshoots or undershoots by a noticeable margin. Two-pass analyses first and allocates second, which is why it is the standard for anything with a hard size limit.
Audio is the line most often forgotten. It is small relative to video and it is not zero, and a target computed without it will be missed by exactly its size.
Container and codec are separate decisions that get confused constantly. An MP4 is a container; H.264, HEVC and AV1 are codecs that can live inside it. Two files with the same extension can therefore need entirely different software to play, which is why "it is an MP4" says nothing about whether a given device will open it.
What does frame rate do?
Multiplies the data linearly, and changes the arithmetic of duration. Duration is frame count divided by frame rate: 1,440 frames at 24 fps is exactly 60 seconds, and the same frames at 25 fps run 57.6 seconds.
That four per cent is the classic PAL speedup — conforming 24 fps film to 25 fps for European broadcast simply plays it faster, shortening a two-hour film by nearly five minutes and raising the pitch of the audio by a semitone unless it is corrected.
It is also why timecode and real time diverge on anything shot at 29.97 rather than 30, and why drop-frame timecode exists to reconcile them.
Questions people ask
Is a higher bitrate always better? Up to the point where the encoder has enough data to represent the source, after which it is wasted storage. Where that point sits depends on the content — a static interview needs far less than fast motion.
Why does my card hold less than advertised? Decimal versus binary gigabytes, plus filesystem overhead. Expect roughly 7 per cent less usable space than the number on the label.
Does variable bitrate change the calculation? It makes the average the figure that matters. A VBR file with an average of 8 Mbps produces the same size as a CBR file at 8 Mbps; the difference is where the bits are spent.
What is a safe recording bitrate? More than you think you need, because it is the only decision you cannot revisit. Storage is cheap and re-shooting is not.
Does audio bitrate need two passes too? No — audio encoders hit their target closely in one pass, which is why the video allocation is the part that needs analysis.
Why does streaming look worse than the same file locally? Because adaptive streaming drops the bitrate when the connection cannot sustain it, keeping the resolution label and reducing the data behind it. The player reports 4K throughout while delivering considerably less.
Bitrate is what was stored; resolution is what was requested. The video bitrate calculator sizes the encode, file size from bitrate and bitrate from file size work the two directions of a target, the recording time calculator turns a card into minutes, and the frame rate calculator covers the duration and timecode arithmetic.