While decades of research have deepened our understanding of time perception, the perception of (manipulated) video speed has been relatively underexplored but is gaining interest with recent technological advances. This study systematically investigated human perception of "video speed," "clip duration" and "original duration" across slow motion, original speed and time lapse. Results showed that participants consistently underestimated video speed in time lapse and overestimated it in slow motion, suggesting a tendency toward an internal perceptual standard. A similar pattern emerged for clip duration with videos being generally overestimated in their duration when played in accelerated speed, and underestimated in slow motion. For original duration estimations, this pattern was reversed. Surprisingly, while estimations of clip and original duration were highly correlated in all video speed conditions, both measures were uncorrelated with estimated video speed. However, an exploratory analysis suggested that the distorted perception of original duration may, at least in parts, still be rooted in participants relying on their own (but biased) interpretations of clip duration and video speed. Most importantly, these results reveal distortions in perception caused by altered video speeds, suggesting caution when employing these video techniques for judgments and decision-making. The findings provide foundation for further research, investigating the cognitive mechanisms of human video speed perception.