PerFrame

Perframe, usually written as per-frame, is a term used in video, animation, and graphics to denote data, decisions, or processing that applies to a single frame within a sequence. It contrasts with per-shot or per-scene processing, which spans multiple frames. In encoding and processing, per-frame strategies can vary on a frame-by-frame basis, enabling dynamic adaptation to content complexity or display conditions. For example, per-frame quality metrics, quantization parameters, and reference-frame selection can be adjusted for each frame. In rendering, per-frame computations may include shading, lighting, and post-processing effects evaluated for every frame, allowing temporal changes to be reflected in the animation.

Per-frame metadata includes timestamps, frame dimensions, color space information, and compression settings that may differ from one frame to the next. In streaming and broadcast, per-frame adjustments can implement fine-grained rate control, error resilience, or dynamic range handling. In computer vision and analytics, per-frame analysis involves feature extraction, object detection, and motion estimation applied independently to every frame, followed by temporal tracking across frames.

Challenges with per-frame approaches include increased computational load, memory usage, and potential temporal artifacts if frame-to-frame changes are not coherently constrained. Benefits include greater adaptability and precision for time-varying content.

See also: frame, frame rate, I-frame, P-frame, B-frame, GOP, motion vector.