INSIDE

New Apple AI model edits images based on natural language input

Tim Hardwick

Apple researchers have released a new open-source artificial intelligence model that can edit images based on user instructions in natural language (via VentureBeat).

MacRumors image created with DALL·E
Name “MGIE”, which stands for MLLM – Guided image editing uses multimodal large language models (MLLM) to interpret user queries and perform pixel-level manipulations.

The model is capable of editing various aspects of images. Global photo enhancements can include brightness, contrast, or sharpening, as well as artistic effects such as sketching. Local editing can change the shape, size, color or texture of certain areas or objects in an image, and Photoshop-style modifications can include cropping, resizing, rotating and adding filters, or even changing the background and blending images.

Custom request to a photo of pizza might be to “make it healthier.” Using common sense, the model can add vegetable toppings such as tomatoes and herbs. A global optimization request could take the form of “add contrast to simulate more light,” and a Photoshop-style modification could be made by asking the model to remove people from the background of the photo, shifting the focus of the image to the subject's facial expression.

Apple, in collaboration with researchers at the University of California, created MGIE, which was presented in a paper at the International Conference on Learning Representations (ICLR) 2024. The model is available on GitHub and includes code, data, and pre-trained models.

This is Apple's second breakthrough in artificial intelligence research in recent months. In late December, Apple announced that it had made progress in deploying large language models (LLM) on the iPhone and other Apple devices with limited memory by inventing an innovative technique for using flash memory.

Over the past few months, Apple has been testing an Apple competitor GPT, which could compete with ChatGPT. According to Bloomberg's Mark Gurman, working on artificial intelligence is a priority for Apple as the company develops the Ajax platform for large language models.

Both The Information newspaper and analyst Jeff Pu say Apple will have some kind of The generative AI feature will be available on iPhones and iPads around the end of 2024, when iOS 18 releases. According to Gurman, iOS 18 includes an enhanced version of Siri with a generative AI feature similar to ChatGPT, and could potentially be the “biggest” software update yet software in the history of the iPhone.

Tag: Apple GPT Guide[ 60 comments ]

Leave a Reply

Your email address will not be published. Required fields are marked *