You AirDropped a 45-second clip from your iPhone to your Mac and the file is 780 MB. You uploaded a 20-minute Zoom recording and it took ten minutes to send. You emptied your GoPro's SD card and the folder is 340 GB of footage from one weekend. All three files feel disproportionate to what is on screen. They are not - they are what current video actually costs, and the formula is exact.

#The formula

Video file size in bytes is very close to bitrate (bits per second) × duration (seconds) / 8. The bytes for audio, container overhead, and metadata are usually 1-3% of the total; the video track is where the file lives. If you know two of those three variables, you can predict the third within a few percent.

Where it gets interesting is what pushes the bitrate up. Bitrate is set by the encoder at record time to hit a target visual quality, and that target scales with resolution, frame rate, bit depth, dynamic range, and codec efficiency. Move any of them and the bitrate moves proportionally.

#Resolution: quadratic scaling

Resolution is pixel count, and pixel count is width × height. Going from 1080p (2.07 million pixels) to 4K (8.29 million pixels) is a 4x increase in the number of pixels the encoder has to describe. All else equal, that means a 4x increase in the visually-transparent bitrate. A 1080p H.264 stream at 8 Mbps looks like a 4K H.264 stream at around 32 Mbps.

This is why 4K files feel disproportionately large: they are, correctly, four times the size of the "same" clip at 1080p. And why 8K, when it eventually happens, will be four times the size of 4K again.

#Frame rate: linear

Twice as many frames per second = twice as much data to encode. 60 fps is 2x the bitrate of 30 fps at the same visual quality; 120 fps is 4x. There is a small efficiency gain because consecutive frames are more similar at higher rates (the encoder gets better predictions), but it is small - call it 10-20%. High-frame-rate slow-motion footage is where GoPro and iPhone ProRes files start hitting hundreds of megabits per second.

#Bit depth: SDR versus HDR / 10-bit

An SDR video stores each pixel as 8 bits per channel (256 gradations per colour). A 10-bit HDR video stores 10 bits per channel (1,024 gradations). That is 25% more data per pixel before compression, and because the encoder has to preserve finer gradations to avoid banding, the actual bitrate increase is closer to 35-50%. iPhone 4K HDR Dolby Vision at 60 fps sits around 470 Mbps peak - 30 GB per hour.

#Codec efficiency

Newer codecs need fewer bits to hit the same visual quality. HEVC saves roughly 40% versus H.264. AV1 saves another 20-30% versus HEVC. Your iPhone's default "High Efficiency" mode records HEVC precisely to shrink files that would otherwise be H.264-huge. See the codec comparison for the trade-offs.

#Real numbers from real cameras

All numbers are approximate peak bitrates in typical recording modes:

  • iPhone 15 Pro, 1080p 30 fps H.264: about 25 Mbps. That's 190 MB per minute, 11 GB per hour.
  • iPhone 15 Pro, 4K 60 fps HEVC: about 100 Mbps. About 750 MB per minute, 45 GB per hour.
  • iPhone 15 Pro, 4K 60 fps ProRes (external SSD required): about 1,700 Mbps. About 12.5 GB per minute, 750 GB per hour.
  • GoPro Hero 12, 5.3K 60 fps HEVC: about 120 Mbps. About 900 MB per minute.
  • Sony A7 IV, 4K 60 fps XAVC HS (HEVC): about 200 Mbps. About 1.5 GB per minute.
  • DJI Mavic 3, 4K 60 fps H.264: about 200 Mbps.
  • Screen recording, macOS Cmd-Shift-5 at 1080p 30 fps: about 8-20 Mbps depending on scene complexity.
  • Zoom recording, 1080p at 30 fps H.264: about 1-3 Mbps (heavily compressed for real-time streaming).

#Audio's contribution

Almost negligible. Stereo AAC at 192 kbps is 24 kilobytes per second, or 1.4 MB per minute. Even a 5.1 AC-3 track at 640 kbps is under 5 MB per minute. On a 4K HDR video pulling 750 MB per minute for the video track, audio adds less than 1%.

#Why "compress it" is a legitimate ask

Cameras record with bitrates tuned for professional post-production headroom: the encoder writes more data than the viewer's eyes will ever use, so an editor can grade the colour and re-encode without visible loss. For distribution, that headroom is waste. A 45 MB per minute iPhone 4K HEVC clip can be re-encoded to 15 MB per minute of HEVC at the same visible quality on a phone screen. That is not "losing quality" - it is dropping headroom the viewer never sees.

The compression article walks through how to do this without going too far.

#When the wrapper is the problem, not the size

If your file is big and playing wrong on a target device, size is often not the reason. An MKV that will not play on iPhone is a container problem, not a size problem - the same bytes in an MP4 wrapper play instantly. Remuxing keeps the file the same size and fixes the playback question in seconds. Only reach for the compression toolkit when the file really is too big for the destination (a 3 GB clip to a service with a 100 MB upload cap, a 40 GB archive against a full disk).

#The takeaway

Video files are big because pixels × frames × bit depth × dynamic range is a genuinely huge amount of information, and cameras write it at bitrates designed for editing headroom. The math is exact: bitrate times duration equals bytes. Change the resolution or the codec or the bit depth and the file size changes with it. The choice of how much to compress is a viewer question, not a source question - pick a bitrate that looks right on the device the video will be watched on, and stop there.