Video quality measurement is the practice of turning the question of how good a compressed or delivered video looks into a reproducible, scalable, comparable number, rather than an opinion that shifts with the viewer, the screen, and the room. It is done in two ways: subjective testing, where a panel of viewers rates clips under controlled conditions to yield a Mean Opinion Score, and objective metrics such as PSNR, SSIM, and VMAF, which compute a score in software in seconds. The key catch is that the target is perceived quality, not pixel arithmetic, so every automatic score is a proxy with blind spots, and a single pooled number can hide a bad scene that the average smooths away. A measurement program does four jobs: compare encoders, set quality targets, catch regressions, and prove delivered quality. It underpins related concepts such as QoE, the reference setups, and the objective-versus-subjective split.
