Meta has recently expanded its Expressive Media Universe (EMU) technology to include two new tools that focus on image editing and video generation. The EMU technology was initially presented at the Connect event in September, where it introduced a tool for creating ‘stickers’ to be used in messaging applications.
The first tool, EMU edit, allows for precise image editing from text instructions. This generative AI technology follows instructions to modify pixels in the input image without affecting the entire image. It also incorporates computer vision tasks as instructions for image generation models, improving precision and control in the editing process.
The second tool, EMU video, facilitates the generation of videos from a text description through a unified architecture that responds to different types of inputs. It uses diffusion models to generate videos, providing a streamlined process for video generation tasks.
These new tools represent a significant step forward in image editing and video generation for Meta, building on the capabilities of its founding content generation model, EMU. With advanced features for precise and controlled editing, as well as efficient video generation from text descriptions, these tools are poised to revolutionize media creation and consumption.