Automated QC is the set of machine checks that sit at fixed points in a video pipeline and turn a measurement into a decision, pass, warn, or block, without a human watching every frame. A metric is just an input; QC is the surrounding machinery that compares it to a threshold and acts. There are three natural checkpoints: ingest validation confirms the source conforms to spec, a post-encode quality gate scores the encode against the source (a full-reference measurement, typically VMAF), and a pre-publish check validates the packaged deliverable. Checks span layers from cheap structural reads to decoding the actual frames for black-frame, freeze, blockiness, and loudness detection, plus modern machine-learning detectors for perceptual artifacts. Run the cheap checks first and fail fast. The key catch is the design choice between a hard gate that blocks and a soft gate that only warns, and tuning for recall without flooding humans with false alarms, since real defects are rare. The point is to spend human judgment only where it is needed.
The handbook · 7 volumes
Fora Soft Video Engineering Handbook
Automated QC only pays off when each check has a threshold and an action that match where it runs: ingest, post-encode or pre-publish. We wrote seven volumes covering the whole video stack, and for automated QC start with Volume 1: perceptual metrics, compression artifacts and encoder settings, with Volume 3 on player-side QoE monitoring and Volume 6 on AI models for video. Written by Nikolay Sapunov, CEO at Fora Soft.
Still eyeballing encodes before publish because automated QC misses real problems?
Send us your pipeline and recent QC escapes: we map the checks, thresholds and gate points for automated QC and estimate the review time it saves. 250+ video products shipped since 2005.
