Lumiere is a new AI model for video creation unveiled by Google researchers. It provides unmatched realism and creative flexibility, generating the whole video simultaneously to keep everything consistent.
Lumiere uses a cutting-edge “space-time” neural network to produce 5-second video clips in one go, unlike other models that generate separate keyframes. This method ensures a smooth video experience, reducing the ‘choppiness’ often found in other videos. Lumiere can make videos from text prompts or animate parts of still images, similar to Pika Labs and Runway. The demo also features video inpainting and style cloning.
The recent progress in AI video creation has been amazing, and we believe the industry will keep surprising us throughout 2024. The research that was hidden is now starting to come to light after all the drama in Hollywood last year.
Check out the full demo here.
