全球知名人工智能研究机构Stability AI近日宣布发布了其第三代图像生成模型——Stable Diffusion 3。这款最新模型在文字渲染能力、多主题提示功能以及图像质量上实现了显著提升，预示着人工智能在图像生成领域的又一重大突破。
据官方展示的样例图片，Stable Diffusion 3的生成效果已经可以与Midjourney等业界领先模型相媲美，甚至在某些方面展现出超越的潜力。在技术层面，该模型提供了更广泛的参数选择范围，从800兆字节（M）到80亿字节（B），为用户提供了更灵活的定制化选项。
尽管详细的技術报告尚未公开，Stability AI已经透露，Stable Diffusion 3的核心技术融合了扩散型Transformer架构（DiT）和Flow matching。这种创新的结合方式有望在保持高图像质量的同时，提升模型的效率和多样性。
Stability AI的这一进展，不仅将影响到艺术创作、设计行业，也将对广告、电影特效等领域产生深远影响，推动人工智能在视觉内容生成方面的应用达到新的高度。随着更多技术细节的披露，Stable Diffusion 3有望引领一场图像生成技术的革新风暴。
来源：量子位
英语如下：
Title: “Stability AI Launches Stable Diffusion 3: A New Milestone in Image Generation Technology”
Keywords: Stability AI, Stable Diffusion 3, Image Generation
Content:
Stability AI, a renowned global AI research institution, has recently announced the release of its third-generation image generation model, Stable Diffusion 3, marking another significant leap in technological innovation.
The latest model boasts substantial improvements in text-to-image rendering capabilities, multi-themed prompts, and image quality, signifying a major breakthrough in the field of AI-generated imagery. As per the samples showcased by the official source, Stable Diffusion 3’s output is comparable to industry leaders like Midjourney, with potential advantages in certain aspects.
Technically, the model offers a broader range of parameter options, from 800 megabytes (MB) to 8 billion bytes (B), enabling users to have more flexible customization choices.
While detailed technical reports are yet to be disclosed, Stability AI has hinted that Stable Diffusion 3’s core technology integrates Diffusion-based Transformer architectures (DiT) with Flow matching. This innovative combination aims to enhance the model’s efficiency and diversity while maintaining high image quality.
The progress made by Stability AI is poised to impact not only the realms of artistic creation and design but also advertising and film special effects, propelling AI-driven visual content generation to new heights. As more technical details come to light, Stable Diffusion 3 is expected to spearhead a revolution in image generation technology.
Source: Quantum Bit
【来源】https://www.qbitai.com/2024/02/122881.html
Views: 1