The tech company Apple has made headlines by developing an open-source image editing AI model called MLLM-Guided Image Editing (MGIE), revolutionizing the way users can manipulate images. This innovative technology uses multimodal large language models to interpret text-based commands when editing photos. Developed in conjunction with researchers from the University of California, Santa Barbara, the MGIE can interpret vague text prompts into detailed, clear instructions for the photo editor itself to follow. Not only can MGIE make significant changes to images, but it’s also capable of cropping, resizing, and rotating photos, as well as improving brightness, contrast, and color balance.
Furthermore, the model has the capability to edit specific focal areas within a photo and can adjust elements such as hair, eyes, and clothing on a person or remove background features. Although Apple has shared the model through GitHub and even hosts a demo on Hugging Face Spaces, it’s still unclear whether the company intends to use the technology in future products. Despite this, MGIE has showcased potential to enable text-based image editing that is more efficient and detailed than ever before, emphasizing Apple’s role in contributing to artificial intelligence development within the industry.
