Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Google's New AI Text-to-Video Tool Is Fun to Look At. But What Next?

Jan 25, 2024 - cnet.com
Google's Research arm has unveiled Lumiere, an AI-based video generation tool that can create high-quality videos from simple text prompts. The tool can generate text-to-video and image-to-video content, animate famous paintings, and fill in missing visuals within a video clip. It can also change the style of the subject completely using prompts like "made of colorful toy bricks" or "made of flowers." Lumiere uses a space-time diffusion model, processing videos in multiple space-time scales to create realistic and coherent motion.

However, it's unclear when or if the tool will be available to anyone outside Google. While Lumiere's capabilities hold promise, some of the generated images, particularly of animals, betray their CGI origins. Despite this, the tool could make video editing more accessible to those without a background in graphic design. The tool's name, Lumiere, is French for "light."

Key takeaways:

  • Google's Research arm has developed an AI-based video generation tool called Lumiere, which can create high-quality videos from simple text prompts.
  • Lumiere's capabilities include text-to-video and image-to-video generation, stylized generation, and the ability to fill in missing visuals within a video clip.
  • The tool uses a "space-time diffusion model" to generate videos, focusing on smooth, flowing motion for every frame, which is superior to existing models that synthesize distant keyframes followed by temporal super-resolution.
  • While Lumiere holds a lot of promise, some of the generated images, particularly of animals, betray their CGI origins, indicating there is still room for improvement.
View Full Article

Comments (0)

Be the first to comment!