Glossary
Every term you meet while making a video smaller, defined in plain English and with the number attached where there is one. No entry here is longer than it needs to be.
A
- Aspect RatioThe proportions of the frame. Mismatching it is how you get black bars, and black bars are data you are paying to store.
- Audio BitrateThe share of the budget the soundtrack takes. Usually small enough to ignore — except on short clips, where it can be a third of the file.
- AV1The most efficient codec in wide use and the only royalty-free one. Around 30% smaller than H.265, at an encoding cost that is no longer prohibitive.
B
- BandingStepped bands where a gradient should be smooth. Usually a bit-depth problem masquerading as a bitrate problem, which is why raising the bitrate often does not fix it.
- Base64 EncodingRepresenting binary data as text, at a cost of about a third more bytes. The reason a 25 MB email attachment is really an 18 MB file.
- Bit DepthHow many distinct values each colour channel can take. 8-bit gives 256 per channel, which is fewer than smooth gradients need.
- BitrateHow many bits of data a video spends per second. It is the lever that decides file size, and the arithmetic linking the two has exactly one answer.
C
- CBR (Constant Bitrate)Spending the same data rate every second regardless of what is on screen. Wasteful for files, necessary for live streaming.
- Chroma Subsampling (4:2:0)Storing colour at lower resolution than brightness, because eyes care less about it. Free savings for camera footage, and a problem for coloured text.
- CodecThe algorithm that encodes and decodes the video. Distinct from the container, and confusing the two is behind most 'why will this not play' problems.
- Compression ArtifactsThe visible damage lossy compression leaves: blocking, mosquito noise, smearing. Knowing their shapes tells you which setting went too far.
- Container (MP4, MOV, MKV, WebM)The wrapper that holds video, audio and timing data together. It decides what a file is called, not how well it compresses.
- CRF (Constant Rate Factor)A quality target for the encoder, not a size target. Lower means better and bigger; the scale is inverted, and the right value depends on what you filmed.
F
- FFmpegThe open-source toolkit nearly every video tool is built on, including the ones that do not mention it. Worth knowing exists, even if you never type a command.
- File Size LimitA platform cap on upload size. Worth knowing exactly what it measures, because the number on the label is often not the number you can send.
- Frame Rate (fps)Frames per second. Tempting to halve for an easy saving, and usually the wrong lever — the cost lands on motion, where viewers notice most.
G
- Generation LossThe quality that compounds away each time a video is re-encoded. The reason a clip forwarded through three apps looks like it has been through three apps.
- GOP (Group of Pictures)The run of frames from one keyframe to the next. Its length is a direct trade between file size and how precisely the video can be seeked or cut.
H
- H.264 / AVCThe most compatible video codec there is, and the default worth keeping. Not the most efficient — it is a 2003 standard — but the one that plays everywhere.
- H.265 / HEVCRoughly half the bitrate of H.264 at the same quality, with a compatibility cost and a licensing history. Best when you control the playback device.
- Hardware Acceleration (NVENC, QSV, VideoToolbox)Encoding on a dedicated chip instead of the CPU. Far faster, and measurably less efficient at the same quality.
K
L
- Lossless CompressionCompression that reconstructs the original exactly. Real, useful for archival and editing, and not how you fit a video under an upload limit.
- Lossy CompressionDiscarding information permanently to save space. Every practical video compressor is lossy; the question is only whether you can see what went.
M
P
R
- RemuxingRepackaging the same encoded streams into a different container. Instant, lossless, and the right answer far more often than people reach for it.
- ResolutionThe pixel dimensions of each frame. The lever with the best ratio of size saved to quality lost, and the most commonly over-provisioned setting in video.
S
T
- TranscodingDecoding video and re-encoding it, usually into a different codec or bitrate. Always lossy, and something the platforms do to your file whether you like it or not.
- Two-Pass EncodingAnalysing the video once before encoding it, so a fixed bitrate is spent where it is needed. The reason a target-size encode lands accurately.
V
- VBR (Variable Bitrate)Letting the data rate rise and fall with how demanding the footage is. What every sensible file encode does, including CRF.
- VMAFNetflix's open perceptual quality metric, scored 0-100 against the original. It is what every quality claim on this site is measured with.
- VP9Google's royalty-free predecessor to AV1, roughly on par with H.265. Still the pragmatic web codec on devices too old for AV1.
W
Looking for the measurements rather than the definitions? Read the guides.