SAN FRANCISCO, CA – Apple has teamed up with researchers at the University of California at Santa Barbara to develop a new image editing tool that utilizes text-based instructions. While this tool, known as Multimodal Large Language Model Guided Image Editing (MGIE), has not been officially released, a demo is available for those interested in trying it out.
In a market filled with AI image editors, MGIE stands out for its innovative approach. By incorporating a text-based AI to interpret and enhance user instructions, the tool aims to produce more accurate and descriptive results. The researchers behind MGIE believe that this additional layer of AI interpretation is crucial for instruction-based image editing, leading to a notable improvement in the final output.
Despite the potential of MGIE, it is uncertain whether the tool will ever be officially released by Apple. The company has not provided any official statement regarding the future of MGIE. However, an open-source version of the software has been published on GitHub, allowing those familiar with the platform to experiment with MGIE.
In a practical test conducted at the Gizmodo office, the MGIE tool demonstrated both impressive and bizarre results. While the tool excelled in some cases, it also presented limitations commonly associated with AI image generators. As with any AI technology, familiarity with the subject matter yielded better results, suggesting that MGIE may perform better with recognizable subjects.
It is worth noting that this collaboration between Apple and the University of California at Santa Barbara serves as a prime example of the ongoing advancements in AI technology. The development and potential release of MGIE further highlight the intersection of artificial intelligence and image editing, offering a glimpse into the future of digital content creation.