The clip opens on a relaxed man looking at his phone in a yellow duck float. The float moves beyond the pool corner and tilts steeply as he leans over the drop, hangs an arm down and struggles to regain balance. Splashes appear beside the glass wall before the float returns to the pool. Neither the man nor the float falls. The actual tilt differs from the public prompt's instruction to keep the ring horizontal. The opening composition places water on the right and open air on the left, with the pool edge visible between them. This gives each timed action a stable spatial reference. One man, one float and a fixed wide view reduce competing movement, while the final return to the water gives the sequence a readable ending. The supplied output also shows a limitation: asking for an unchanged horizontal ring does not guarantee that orientation.
- Opening frame
- One man and a yellow duck float beside a skyscraper pool edge
- Delivered dimensions
- 720 × 1264
- Delivered duration
- 15.041667s
Read the full case and original prompt →At street level, pedestrians run toward the camera while a large tan dog moves between tall buildings and yellow taxis. A hard cut moves inside an office, where the dog's face appears outside the window and turns sideways. Another hard cut returns to a low street view as the dog lifts its muzzle toward small helicopters overhead. The dog remains gentle-looking; the comedy comes from its scale and the surrounding reactions. The opening image establishes the dog's tan coat, floppy ears and dark muzzle. The video prompt repeats those identifying features across three timed views. Each viewpoint adds a separate scale cue: people and taxis at street level, upper-floor windows beside the head, and helicopters above the muzzle. Explicit cut instructions let one generation change perspective without adding an unseen reference image for each shot.
- Opening frame
- One giant tan dog, pedestrians and yellow taxis in a city avenue
- Delivered dimensions
- 848 × 1072
- Delivered duration
- 10.041667s
Read the full case and original prompt →In a factory hall, a large excavator bucket covered in rust and dirt hangs from crane chains above a yellow tank of foaming green water. The crane lowers it teeth first until it disappears completely under the bubbling surface, leaving only the chains visible. The crane then lifts the same bucket out, now clean silver metal inside and out, with its open mouth facing the camera. Water streams from the bucket and its teeth back into the tank as it hangs still at the end. Each reference image fixes one state of the same shot: the rusty excavator bucket above the tank, the bucket hidden under the water, and the clean bucket above the tank. Making images 2 and 3 from image 1 keeps the tank, chains, crane and camera identical. The video prompt times the dip and the lift, keeps the bucket's open mouth toward the camera, and hides the change under the water, so the clean bucket reads as the same object.
- Reference images
- Three ordered images: rusty excavator bucket, bucket fully submerged, clean excavator bucket
- Delivered dimensions
- 720 × 1280
- Delivered duration
- 10.041667s
Read the full case and original prompt →
