近日，科技巨头苹果公司与美国加州大学圣芭芭拉分校（UCSB）的研究团队携手，推出了一项名为MGIE（MLLM-Guided Image Editing）的开源图片编辑框架，旨在解决传统图像编辑中指令不明确的问题。MGIE巧妙地结合了多模态大模型MLLM，让AI能够理解并生成简洁明了的编辑指令，同时提供与视觉内容紧密相关的引导。
据机器之心报道，MGIE框架通过端到端的训练方式，使得扩散模型能够同步更新，利用潜在目标的想象力执行精确的图像编辑任务。这一创新技术使得AI不仅可以实现Photoshop风格的复杂图像修改，还能进行全局照片的优化以及对局部对象的精准调整，极大地提升了图像编辑的智能性和便利性。
这一开源项目将为开发者和设计师提供强大的工具，推动AI在图像编辑领域的应用进一步发展，同时也为未来的多模态人机交互打开了新的可能。苹果与UCSB的这一合作，再次证明了科技巨头在推动人工智能技术与艺术创作融合方面的决心和实力。MGIE的出现，无疑将为图像编辑行业带来革命性的变化，期待它在未来的应用中带来更多的惊喜。
英语如下：
News Title: “Apple and UCSB Join Forces, Launching MGIE: AI-Powered Image Editing Enters a Photoshop-Level Era!”
Keywords: Apple UCSB collaboration, MGIE framework, image editing revolution
News Content:
In recent times, tech giant Apple has partnered with the research team from the University of California, Santa Barbara (UCSB) to introduce an open-source image editing framework called MGIE (MLLM-Guided Image Editing), pioneering a new trend in AI-assisted image editing.
As reported by Machine之心, MGIE addresses the issue of ambiguous editing instructions in traditional image editing by skillfully integrating Multi-Modal Large Language Models (MLLM). This allows AI to comprehend and generate clear editing commands, offering guidance closely tied to visual content.
The MGIE framework, through end-to-end training, enables diffusion models to update simultaneously, leveraging the imagination of the latent target for precise image editing tasks. This innovation not only allows for complex Photoshop-like modifications but also facilitates global photo enhancements and precise adjustments to local objects, significantly enhancing the intelligence and convenience of image editing.
This open-source project will empower developers and designers with powerful tools, advancing AI applications in the realm of image editing and opening new possibilities for multimodal human-computer interaction. Apple’s collaboration with UCSB underscores the commitment and prowess of tech titans in fusing AI technology with artistic creation. The advent of MGIE is set to bring a revolutionary change to the image editing industry, with anticipation for more delightful surprises in its future applications.
【来源】https://www.jiqizhixin.com/articles/2024-02-05-10
Views: 1