Apple releases an AI model that can edit images based on text-based commands
Apple has released an open source AI model for image editing that uses multimodal large language models to interpret text-based commands.
The model, called MLLM-Guided Image Editing (MGIE), can edit photos based on the text the user types in and can also crop, resize, and rotate photos and improve brightness, contrast, and color balance.
Apple has not yet indicated whether it will incorporate this technology into any of its products. [ more ]
Amazon has unveiled the Titan Image Generator, a generative AI model that can generate images based on text descriptions and customize existing images.
The Titan Image Generator offers a range of image editing capabilities, including automatic editing based on text prompts and the ability to swap out backgrounds. [ more ]
Stability AI tries to stay ahead of the pack with a new image-generating AI model
Stability AI's new model, Stable Cascade, is faster and more powerful than its predecessor, Stable Diffusion.
Stable Cascade offers image generation features like variations of images, increasing resolution, inpainting, outpainting, and canny edge creation. [ more ]
Google's AI Magic Editor won't work on IDs, faces, or bodies
Google's AI-powered Magic Editor cannot alter images of ID cards, receipts, human faces, or body parts.
The software has safety guardrails to avoid editing personally identifiable information and to prevent harmful activities like deepfakes or cyberbullying.
Google is working to improve the guardrails and error messages of the Magic Editor. [ more ]
Kate photo: Why this controversy could be just the beginning
The released image of the Princess of Wales led to speculation due to perceived inconsistencies and possible editing.
The prevalence of edited images, possibly by amateur photographers, underscores the challenge of distinguishing real from manipulated content. [ more ]