Reference
194 terms with definitions, aliases, and cross-references to deep-dive articles. Search by name, filter by chapter, or jump straight to a letter.
No terms match your filters. Try a different keyword or clear filters.
1024 levels per channel. Required for HDR delivery (HDR10, Dolby Vision). Also compresses SDR slightly better in HEVC/AV1 thanks to finer quantization.
#
4096 levels per channel. Used by Dolby Vision mastering, professional production (ProRes 4444 XQ) and some VVC profiles.
Chroma subsampling mode where color is halved both horizontally and vertically. The default for delivery — H.264, HEVC, AV1, all streaming.
Chroma subsampling mode where color is halved horizontally. Standard for broadcast and production codecs (ProRes 422, DNxHR HQ).
Chroma subsampling mode with no color reduction — every pixel has full Cb and Cr. Used in mastering, VFX, screen capture.
256 luminance levels per channel. The historic baseline; sufficient for SDR delivery but visibly bands on gradients and fails for HDR.
A set of pre-encoded renditions at different bitrate/resolution combos, designed to give the player a smooth quality-vs-bandwidth tradeoff.
A
Encoder rate-control mode that targets an average bitrate over the whole clip. Often a sensible default for VOD when CRF isn't applicable.
A subjective-testing protocol where each clip is rated alone on a 5-point absolute scale (excellent…bad).
An asymmetric transform used alongside the DCT in AV1, better matching residuals biased toward one edge of the block.
Use of ML inside the encoder loop — CTU partition prediction, scene-cut detection, perceptual quantization tuning. Distinct from neural codecs.
VVC in-loop filter using Wiener-style classification per region to further reduce residual artifacts after SAO.
Artifacts produced when a signal is under-sampled, e.g. moiré on a textured shirt or stairstepping on a near-horizontal line.
AMD's hardware video acceleration framework. Encodes H.264, HEVC and AV1 (RDNA 3 and later).
A near-optimal entropy code that represents the entire message as a single fractional number — the basis of CABAC and AV1's daala_bool.
A custom silicon design for one purpose — in video, used by YouTube (Argos) and others for transcoding at hyperscale.
The ratio of width to height of the displayed image, e.g. 16:9, 4:3, 21:9 cinemascope. Different from pixel aspect ratio (PAR).
2018 open, royalty-free codec from the Alliance for Open Media. ~30 % more efficient than HEVC; mainstream in 2026 thanks to SVT-AV1 and hardware decode.
Successor to AV1 from the Alliance for Open Media, expected to ship around 2025–2026. Targets ~30 % gain over AV1 with similar royalty-free terms.
Visible stepping on smooth gradients (sky, walls) caused by insufficient bit depth. Mitigated by 10-bit encoding or dithering.
B
A single-number summary of how much bitrate codec A saves over codec B at equal quality. The currency of codec benchmarks.
A frame predicted from both past and future reference frames. The smallest and most efficient — but adds decode latency.
The encoder's decision of how many bits to spend on each frame inside a GOP — typically more on I-frames, fewer on B-frames.
Number of bits stored per pixel component. 8-bit = 256 levels; 10-bit = 1024; 12-bit = 4096. More bits reduce banding and enable HDR.
Average number of bits transmitted per second of video. Higher bitrate = more data = (usually) better quality; the chief lever in delivery.
Generic term for a rectangular region of pixels processed jointly by an encoder — sizes range from 4×4 up to 128×128 in modern codecs.
The 4K/HDR-era wide-gamut color space. Much larger gamut than BT.709; used by UHD broadcast, HDR10, Dolby Vision, modern OTT.
The HD-era color space (1990s) — primaries, white point and YCbCr matrix used by HD broadcast, Blu-ray and most streaming below 4K.
Context-adaptive binary arithmetic coder used by H.264 Main/High and HEVC. ~10–20 % better than CAVLC; the dominant entropy coder of the AVC family.
C
CRF with an upper bitrate cap — quality-driven where possible, hard-limited where simple-scene heuristics would overshoot the channel.
Context-adaptive variable-length entropy coder used in H.264 baseline profile. Cheaper to decode than CABAC but ~10–20 % less efficient.
Rate-control mode that keeps the average bitrate constant in every short window. Required by some broadcast and live-streaming pipelines.
AV1's directional in-loop filter that preserves edges while removing ringing artifacts near high-contrast transitions.
ISO standard for encrypting media so the same encrypted stream can be played by multiple DRMs (Widevine, FairPlay, PlayReady).
The color components of a YCbCr pixel, encoding blue-difference (Cb) and red-difference (Cr). Lower-resolution than luma in 4:2:0.
Storing fewer color (chroma) samples than luma samples — the eye is far more sensitive to brightness, so 4:2:0 halves color data with no perceived loss.
GOP fully decodable on its own — no reference outside its boundaries. Mandatory for clean ABR segmentation.
An MPEG standard that lets the same fMP4 segments serve both HLS and DASH — one set of files, two protocols.
An algorithm and bitstream specification for encoding and decoding video, e.g. H.264, HEVC, AV1. Separate from the container that wraps it.
The full range of colors a system can represent. BT.709 covers ~36 % of CIE 1931, BT.2020 ~76 % — hence "wide gamut" for the latter.
A defined system for representing colors — primaries, white point and transfer function. Examples: BT.709 (HD), BT.2020 (4K/HDR), DCI-P3 (cinema).
Uncompressed size divided by compressed size. 4K @ 60 raw is ~12 Gbps; the same content in HEVC at 20 Mbps is ~600:1.
Video where every frame is spaced by the same time interval. Default and recommended mode for delivery encoding.
A file format that wraps coded video, audio, subtitles and metadata together. Independent of the codec — MP4 can carry H.264 or AV1.
Encoding decisions guided by the content itself — face detection, motion analysis, scene classification — to spend bits where the viewer cares.
Encoding decisions guided by viewing context — device type, network, display capabilities. Extends content-aware encoding to the delivery side.
Quality-targeted rate-control mode in x264/x265/SVT-AV1. You set a quality level (e.g. CRF 23) and the bitrate falls where it will.
HEVC's basic coding unit, up to 64×64, recursively split into smaller CUs via a quadtree — drastically more flexible than H.264's fixed 16×16 MB.
Specialist software house for video, real-time and AI products. Founded 2005. 50 in-house engineers.
By submitting data in this form, you agree with the Personal Data Processing Policy.