What does a text-to-video AI like Sora generate?

A text-to-video AI like Sora generates short video clips directly from a written description, predicting how images should change and move from one frame to the next in order to match what the prompt describes. This lets someone type a simple sentence and receive a moving, often startlingly realistic scene without ever using a camera, though the system still has to essentially imagine countless visual details, like lighting and physics, that aren't explicitly stated in the text.

Printed photo albums are physical products requiring printing and assembly, unrelated to generating digital moving video from text. A faster home internet connection is an internet service upgrade with nothing to do with an AI model producing video content.

Early text-to-video clips often show odd glitches, such as hands with the wrong number of fingers or objects that pass through each other, highlighting how physical realism is still one of the hardest things for these models to get right.