Apple releases open-source AI model that can edit images from natural language instructions
Apple and UC Santa Barbara researchers have released MGIE, an open-source multimodal large language model that enables precise image editing through natural language commands. By interpreting text instructions to perform complex pixel-level manipulations, MGIE represents a significant step in applying MLLMs to visual content creation. This breakthrough offers a practical tool for users to modify images intuitively while serving as a substantial research contribution to the field of artificial intelligence. The release highlights Apple’s accelerating commitment to AI integration across its ecosystem, signaling a shift from its previously cautious approach. As the company develops on-device capabilities and prepares for major software updates, projects like MGIE demonstrate its growing influence in generative AI. This move positions Apple to potentially rival leading competitors and enhances the role of open-source contributions in advancing consumer-facing technologies within its hardware and software infrastructure. This initiative is highly relevant to the open data community as it provides fully accessible code, datasets, and pre-trained models on GitHub. By democratizing access to sophisticated image editing technology, Apple fosters transparency and collaboration among developers and researchers. This open approach encourages innovation, allows for independent verification of results, and supports the broader ecosystem of AI development by removing barriers to entry for those interested in exploring multimodal learning applications.
Source: mobilesyrup.comPublished on 2024-02-08
Related news
- Apple Has a New Open-Source AI Image Editor
- Apple's push towards Siri improvements continues with a text-powered open-source image editing AI model
- Apple releases ‘MGIE’, an AI model for instruction-based image editing - MacTech.com
- New localllm lets you develop gen AI apps locally, without GPUs | Google Cloud Blog