Meta announces video generation AI 'Make A Video', releases videos of flying super dogs and teddy bears drawing self-portraits

Meta's AI research institute `` Meta AI '' announced AI `` Make A Video '' that generates a video simply by entering a character string (prompt) and its results.


Introducing Make-A-Video: An AI system that generates videos from text

In July 2022, Meta AI announced 'Make A Scene', which allows you to create high-quality images based on prompts and simple illustrations.

Meta announces ``Make-A-Scene'' that can generate high-quality images from text descriptions and super-rough illustrations-GIGAZINE

'Make A Video' is based on the Text to Image technology of 'converting prompts into images'. It is said that motion learning uses unlabeled, publicly available datasets, ensuring transparency of research.

In the sample, in addition to videos created based on input prompts, videos created based on images are also available.

For example, if you enter the following image ......

You can get an animation of the ship moving like this.

You can also enter multiple images. For example, if you enter the following two

Then the asteroid will move.

You can also input videos to generate new videos. For example, based on footage of astronauts leisurely spacewalking.

Different videos of similar situations were obtained.

in Software, Posted by logc_nt