Although Apple is not currently a major player in the field of AI, the company’s latest open-source AI model for image editing demonstrates its potential in this area. The model, named MLLM-Guided Image Editing (MGIE), leverages multiple language models (MLLMs) to interpret text-based instructions for image editing. In essence, this tool has the ability to alter images based on user input. While this is not a groundbreaking capability, the project paper acknowledges that “people’s instructions are sometimes too brief to fully convey their intent.” Apple collaborated with researchers from the University of California, Santa Barbara to develop MGIE. MLLMs have the capacity to translate simple or ambiguous words into detailed and explicit instructions that a graphic editor can follow. For instance, if a user wants to make a photo of a pepperoni pizza “healthier,” MLLM can interpret this as “add vegetables” and proceed with editing the image accordingly. AppleIn addition to making significant alterations to images, MGIE is also capable of cropping, resizing, and rotating images, as well as adjusting their brightness, contrast, and color balance, all through text instructions. It can also modify other aspects of an image, for example, the hair, eyes, and clothes of a person, or remove background objects. An online demo of MGIE is available on Hugging Face Spaces. Apple has not yet confirmed whether it intends to incorporate the knowledge gained from this project into a tool or product for inclusion in its offerings.
:max_bytes(150000):strip_icc()/GettyImages-2172043823-bf725998695248fcbe4c803136858770.jpg "Dow Jones Nowadays: Shares Shut Decrease as Traders Digest Fed’s Charge Reduce Choice")
