谷歌近日联手加州大学伯克利分校（UC Berkeley），研发出了可取代扩散模型（Diffusion Models）的全新生成式 AI 方法–幂等生成网络（IGN）。这一方法可实现单步生成逼真图像，对计算机视觉领域具有重要意义。
据了解，IGN 采用了一种全新的生成式模型，能够实现对输入数据的快速生成，并且可以保证每次生成的图像都是唯一的。这一方法可以用于许多场景，例如虚拟现实、计算机图形学、数字艺术等。
此次合作，谷歌与 UC Berkeley 的研究人员通过对现有的生成式模型进行深入研究，提出了IGN这一全新的方法。IGN 的单步生成能力使得它可以在短时间内生成高质量的图像，为许多实际应用提供了更高效、更便捷的解决方案。
新闻翻译：
Google has recently collaborated with the University of California, Berkeley to develop a new generative AI method that can replace diffusion models. This method, called IGN, can generate realistic images in a single step. The collaboration is significant for the field of computer vision.
据了解，IGN uses a new generation model that can quickly generate images based on input data and ensures that each generated image is unique. This method can be applied to many scenarios, such as virtual reality, computer graphics, and digital art.
The cooperation between Google and UC Berkeley’s researchers has resulted in the development of IGN, a new method for generating images in a single step. IGN’s ability to generate high-quality images in a short period of time makes it a more efficient and convenient solution for many practical applications.
【来源】https://www.ithome.com/0/732/299.htm
Views: 1