Apple has just released its open-source AI image generation tool – multimodal large language models-guided image editing or simply MGIE. This is far from complete but it is now available via GitHub and comes complete with instructions in the form of a PDF project paper.
Apple MGIE
It uses text instructions to change and edit images from simple tasks like adjusting contrast, brightness, or white balance to more complex tasks like replacing things in images like adding more vegetables to a pizza.
Other highlights of the model include being able to crop or resize photos, autofill image borders, and change someone’s hair, eyes, and clothes.
The tool is a collaboration between Apple and researchers from the University of California, Santa Barbara. It demonstrates the effectiveness of MGIE in improving automatic metrics and human evaluation.
This is Apple’s first foray into generative AI because of this, it is not expected to be used in a device any time soon. Although, it is likely to be a peek into what’s to come from the company’s AI features.
Sources 1, 2 | Featured Image made with Dall-E
Source: Gadget Pilipinas
0 Comments