Success scenario
Failure scenario
Ground truth
Ground truth
Best performing model: VideoPoet (multiframe)
Second best performing model: Runway Gen 3 (i2v)
