Video
Veo
Google DeepMind
Google models for generating videos with audio from instructions and visual references.
A rainy scene changes when the drops and street noise can be heard. Veo works with that relationship between moving images and sound: Google’s models can generate videos with audio from instructions and visual references. A description begins to resemble a small written scene with its own atmosphere.
That is appealing for imagining the opening of a story, exploring a mood or shaping an initial audiovisual sketch. Camera, action and sound belong to the same brief. Plenty of creative choices remain open: an evocative clip may still need another pace, intention or later edit to fit the complete piece.
