2 new generative AI tools unveiled for image generation

San Francisco: Meta has announced two new generative AI tools called Emu Video and Emu Edit for high-quality video and image creation.
At this year’s Meta Connect, the company announced several new developments, including Emu, its first basic model for image generation.

“With Emu Video, which leverages our Emu model, we present a simple method for text-to-video generation based on the diffusion model. It’s a unified architecture for video creation tasks that can respond to different types of input: text only, image only, and both text and image, Meta explained in a blog post late Thursday.
The team breaks the process into two steps: first, creating optimized images based on text prompts, and then producing optimized videos based on both the text and the generated image.

“This split approach to video generation allows us to efficiently train video generation models,” the company said.

Emu Edit is a new approach that aims to streamline various image manipulation tasks and bring advanced capabilities and precision to image editing.

Emu is capable of performing free-form editing through edit instructions, including local and global editing, removing and adding backgrounds, color and geometry changes, detection and segmentation, and more.

“Unlike many generic AI models today, Emu Edit follows instructions precisely, ensuring that pixels in the input image unrelated to the instructions remain untouched,” Meta said.

While there’s certainly no replacement for professional artists and animators, new technologies like Emu Video, Emu Edit, and their ilk can help people express themselves in new ways.