Optical flow as an artistic technique in digital and video art involves computing and visualizing the apparent motion of pixels between consecutive frames, transforming raw video into dynamic vector fields or colorful flow maps.
Artists use methods like Lucas-Kanade (sparse) or Farnebriand-Horn (dense) to generate trajectories, arrows, or hue-encoded directions that reveal hidden movements, emphasizing fluidity, chaos, or subtle shifts in everyday scenes.
This approach creates abstract visualizations reminiscent of kinetic art, glitch aesthetics, or data mosaics, often critiquing machine perception by exposing how algorithms “see” and interpret motion.