New Apple AI Model Can Edit Images Per Your Instructions
Apple researchers have recently unveiled an innovative open-source artificial intelligence (AI) model, MGIE, that enables users to edit images by simply describing the changes they want to see. This Multimodal large language model (MLLM) Guided Image Editing tool promises to revolutionize the way we approach image modification, offering Photoshop-style adjustments, global photo enhancements, and targeted edits through a user-friendly, instruction-based interface.
Although the current output quality of MGIE has its limitations, its potential applications within Apple’s ecosystem are vast and intriguing. Utilizing sophisticated MLLM AI technology, MGIE excels in parsing natural language instructions and applying corresponding edits to images. It positions itself as a formidable tool in the realm of digital photo editing by improving automatic metrics and human evaluation without compromising on efficiency, as noted in the research publication.
Here’s a glimpse at what MGIE can do:
- Intuitive Editing Instructions: Type in requests like “make the sky more blue,” and MGIE will precisely adjust saturation levels to meet your request.
- Versatile Edits: From basic adjustments like cropping and applying filters to advanced manipulations such as background changes and object blending, MGIE handles a wide range of editing tasks.
- Photo Quality Optimization: Enhance the overall appearance of your photos with adjustments to brightness, contrast, sharpness, and color balance, or add artistic effects for a unique touch.
- Targeted Edits: MGIE can focus on specific image aspects—modifying attributes of faces, clothing, and accessories to achieve the desired look.
The MGIE project is available open-source on GitHub, with an intuitive demo for those eager to test its capabilities with their images. While it’s unclear when this technology will be integrated into Apple products, the combination of AI-powered editing and Apple’s user-centric design philosophy suggests it could soon enhance the way we edit photos on devices like the iPhone.
