A 10-second AI pet comedy: designing cause-and-effect visual storytelling
Can a single salmon carry an entire ten-second story? We tested a simple idea: a giant salmon suddenly slaps a kitten at a supermarket checkout. The surprised kitten collides with a puppy, a conveyor starts moving, milk spills, tomatoes bounce, the dog attempts a rescue, and the salmon returns to its starting position.
The creative structure
The opening is an immediate physical conflict rather than a static character introduction. Each new action produces a consequence, so the scene stays legible without dialogue: fish slap → startled reaction → checkout chaos → puppy intervention → flying fish → loop.
Production
The characters and animation were AI-generated from a visual reference, not real filmed animals. The ten-second source was played through FFplay into a virtual Xvfb display and captured as a new video with audio output recorded through an ALSA PCM plugin. The finished vertical video is 1080×1920 and 24fps, with the fixed on-screen question “ARE WE CUTE?”.
Measurement, not mythology
We will evaluate 1-second retention, average viewing time, completion and sharing against earlier clips. This is a creative experiment, not a claim that AI realism or screen recording improves recommendation eligibility.
Question: What would you change in the first second to make the joke even clearer?
The new hook tests an affirmative/negative reaction, but response volume and reach remain empirical unknowns.
Top comments (0)