Video Bitrate Explained: CBR, VBR and CRF

What bitrate measures

A compressed video stream is a sequence of bits. Its bitrate is how many of them are used, on average, for each second of playback. Video bitrates are usually quoted in megabits per second (Mb/s) and audio bitrates in kilobits per second (kb/s). The total bitrate of a file is the sum of its streams plus a little container overhead.

Bitrate is a measure of quantity, not quality. Two files at 8 Mb/s can look very different if one uses a more efficient codec, has less motion, or was encoded with a slower, more careful preset. Within one codec and one piece of content, though, more bits generally means more detail survives, up to the point where extra bits buy nothing visible.

From bitrate to file size

The arithmetic is simple and worth knowing, because it lets you predict a file's size before you export it:

Add the video and audio bitrates to get the total, in megabits per second.
Multiply by the duration in seconds to get megabits.
Divide by 8 to get megabytes.

A 10-minute video at 8 Mb/s with 128 kb/s audio is about 8.13 × 600 / 8, or roughly 610 MB. For quick estimates, these hourly figures help:

2 Mb/s: About 0.9 GB per hour
5 Mb/s: About 2.25 GB per hour
10 Mb/s: About 4.5 GB per hour
20 Mb/s: About 9 GB per hour
50 Mb/s: About 22.5 GB per hour

These are decimal gigabytes; operating systems that report in binary units show slightly smaller numbers. Audio is usually a small share: 128 kb/s adds under 60 MB per hour. How audio bitrate affects speech specifically is covered in audio bitrate for speech recognition.

Rate control: CBR, VBR and quality targets

Rate control is the encoder's strategy for distributing bits across the video. Three families cover nearly every export dialog.

Constant bitrate (CBR) keeps the data rate close to a fixed value every second. Easy scenes get more bits than they need and hard scenes get fewer. It suits live streaming and broadcast links, where the connection has a fixed capacity, and is rarely the right choice for a file.
Variable bitrate (VBR) aims for an average while letting complex moments take more and simple moments take less, usually within a maximum. Two-pass VBR analyzes the whole video first, then encodes, which hits the target size accurately with better quality than a single pass. It suits files that must fit a size limit.
Quality-targeted modes, such as CRF in x264 and x265, constant quality in many editors, or CQ on some hardware encoders, aim for a consistent visual quality and let the bitrate land wherever it needs to. Size is not predictable in advance, but quality is even across the video and no bits are wasted on easy scenes.

For x264, CRF runs from 0 to 51, where lower means higher quality and larger files; the default is 23 and values from the high teens to the high twenties are commonly used. x265 uses a different scale with a default of 28. Streaming services often combine approaches with a capped quality mode: a CRF target plus a maximum bitrate and buffer size, so quality is consistent but peaks cannot exceed what viewers' connections can handle.

What drives bitrate up

For a given quality, the bitrate an encoder needs depends on several factors:

Resolution. More pixels need more bits, though not proportionally: 4K has four times the pixels of 1080p but usually needs less than four times the bitrate for similar perceived quality.
Frame rate. Doubling frames adds data, but consecutive frames at 60 fps are more alike, so the cost is well under double.
Motion and detail. Fast camera moves, water, foliage, confetti, crowds and film grain are expensive. A talking head against a plain wall or a slide presentation is cheap.
Noise. Sensor noise in low light is random detail the encoder tries to preserve. Denoising before encoding can save a surprising amount of bitrate.
Codec. Newer codecs such as HEVC, VP9 and AV1 generally reach the same quality at lower bitrates than H.264. The video codec explainer describes how.
Encoder preset. Slower presets search harder for efficient ways to describe each frame, giving smaller files at the same quality.
Keyframe interval. Frequent keyframes cost extra bits, as explained in keyframes and GOP structure.

A useful normalized measure is bits per pixel: bitrate divided by width times height times frame rate. It lets you compare settings across resolutions, though acceptable values still depend on codec and content.

Choosing export settings

Master or archive copy: High quality: a low CRF, a high bitrate or an editing codec. Storage is cheaper than re-shooting
Upload to a video platform: High quality, because platforms re-encode everything. Follow the platform's current published upload recommendations
Video played directly from your own site: Quality mode with a maximum bitrate, or two-pass VBR, at a level your viewers' connections can sustain
File that must fit a size limit: Two-pass VBR with the target computed from the limit and the duration
Screen recordings and slides: Quality mode; low bitrates are enough because most of the picture is static
Live streaming: CBR or capped VBR at the platform's recommended rate for your resolution and frame rate

To compute a size-limited target, divide the limit in megabits by the duration in seconds and subtract the audio bitrate, leaving a small margin for overhead. A 500 MB limit for a 20-minute video gives 4,000 megabits over 1,200 seconds, about 3.3 Mb/s total, so roughly 3.1 Mb/s for video with 128 kb/s audio. Fixes for specific upload limits on email, chat apps and course platforms are collected in video file too large.

Example: an oversized product demo

A hypothetical 12-minute 1080p product demo, mostly screen capture with a small camera inset, is exported at a fixed 50 Mb/s and comes out at about 4.5 GB. Re-exporting from the editor in a quality mode around the default CRF produces a file a small fraction of that size that looks the same in playback, because most of each frame is static. The 50 Mb/s setting was sized for fast-moving camera footage, not for a screen recording.

Bitrate mistakes and trade-offs

Comparing bitrates across codecs. 6 Mb/s HEVC and 6 Mb/s H.264 are not the same quality.
Raising the bitrate on a poor source. Extra bits faithfully preserve blur, noise and compression artifacts from earlier generations; they do not add detail.
Re-encoding a file to make it smaller, then re-encoding the result again. Each generation adds loss; go back to the best source.
Using CBR for files. It wastes space on easy scenes and starves the hard ones.
Starving dark scenes and gradients. Low bitrates show up first as blocking in shadows and banding in skies and smooth walls.
Choosing bitrate by resolution alone. A static 4K slide deck needs far less than a 1080p sports clip.

mydubly and bitrate

The bitrate of your video has almost no effect on what mydubly uploads. Only the audio is sent, decoded in your browser, downmixed to mono at 16 kHz and encoded to Opus at 32 kb/s in chunks of about 30 seconds, roughly 120 KB each, for a transcript; a dub uses 24 kHz at about 48 kb/s, roughly 180 KB per chunk. A one-hour video at 20 Mb/s is about 9 GB on disk, but its audio upload is around 14 MB, or about 22 MB for a dub. The video picture never leaves your device. Why that matters is the subject of reducing upload size for video translation.

For a dubbed video, the original picture is copied packet by packet without re-encoding, so its bitrate, quality and size are essentially unchanged; the dubbed AAC voice track replaces the original audio. If you need a smaller file, compress the source in an editor first or compress the result afterwards; mydubly does not change video bitrate. On devices with limited memory, very large files may be offered as the voice track alone, which you can then combine with the video in an editor. mydubly accepts files up to 2 hours long, including MP4 and MOV recordings.

Next step: check your last export

Open your most recent export in a media inspector, note its video bitrate, resolution, frame rate and duration, and compare its size with the arithmetic above. If the bitrate looks high for simple content, try a quality-mode export and compare the two side by side. Once you have a final file at the size you want, the mydubly video translator keeps that picture exactly as it is.

Frequently asked questions

What is a good bitrate for 1080p video?

It depends on the codec, frame rate and content, so a single number is misleading. A static screen recording can look clean at a few megabits per second, while fast sport may need several times more. Use a quality-targeted mode for files, and the platform's published recommendations for uploads and live streams.

Is higher bitrate always better quality?

Only up to a point, and only within the same codec and source. Beyond a certain level, extra bits produce no visible improvement, just a bigger file. A higher bitrate also cannot restore detail lost in an earlier encode.

Should I use CBR or VBR for exporting?

For files, VBR or a quality-targeted mode is almost always better, because bits go where the picture needs them. CBR suits live streaming and fixed-capacity links. Two-pass VBR is the choice when a file must hit a specific size.

What does CRF mean?

Constant rate factor, a quality-targeted mode in x264, x265 and some other encoders. You choose a quality level and the encoder varies the bitrate to keep quality consistent. Lower values mean higher quality and larger files.

How do I calculate video file size from bitrate?

Add the video and audio bitrates in megabits per second, multiply by the duration in seconds and divide by 8 to get megabytes. For example, 5 Mb/s total for 30 minutes is 5 × 1,800 / 8, about 1,125 MB. Container overhead adds a little on top.

Does lowering the resolution reduce bitrate?

It reduces the bitrate needed for a given quality, so with quality-targeted encoding the file gets smaller. With a fixed bitrate, a lower resolution simply gets more bits per pixel. Downscaling is a common way to fit a size limit with fewer visible artifacts than starving a high-resolution encode.