**Stability AI 推出 Stable Diffusion 3，引领文生图技术新高度**
伦敦——全球人工智能研究机构 Stability AI 近日公开了其最新研究成果——Stable Diffusion 3 文生图模型的详细论文，进一步揭示了这一技术背后的创新理念。据报告，Stable Diffusion 3 在文本到图像的生成领域取得了重大突破，尤其是在排版和对提示的精准响应方面，超越了当前的行业标杆，如 DALL·E 3、Midjourney v6 和 Ideogram v1。
Stable Diffusion 3 的核心是全新的多模态扩散Transformer（MMDiT）架构，这一架构允许模型独立处理图像和语言的表示，从而提升了模型在理解和处理文本时的精度，同时也增强了对拼写错误的容忍度。这一改进对于提高生成图像的质量和与输入文本的对应度具有重大意义。
据 Stability AI 的研究人员表示，Stable Diffusion 3 在人类偏好评估中表现出色，其生成的图像不仅在视觉效果上令人印象深刻，而且在遵循用户输入的提示方面表现出更高的准确性。这一进展预示着文生图技术在内容创作、设计辅助和视觉传达等领域将有更广泛的应用前景。
Stability AI 的这一创新成果有望重塑文本到图像生成的行业标准，为人工智能与艺术、设计的融合开辟新的道路。随着技术的不断进步，未来我们有望看到更多由 AI 创作的精美图像，同时，这也对新闻报道、广告设计和创意产业带来了无限想象空间。
英语如下：
**News Title:** “Stability AI Launches Stable Diffusion 3: A New Benchmark in Text-to-Image Generation超越 DALL·E 3”
**Keywords:** Stability AI, Stable Diffusion 3, Text-to-Image model
**News Content:**
**Stability AI Raises the Bar with Stable Diffusion 3, a Pioneering Text-to-Image Technology**
London – Global AI research institution Stability AI has recently disclosed the details of its latest research, the Stable Diffusion 3 text-to-image model, shedding light on the innovative concepts behind the technology. According to reports, Stable Diffusion 3 has made significant strides in text-to-image generation, outperforming current industry benchmarks such as DALL·E 3, Midjourney v6, and Ideogram v1, particularly in layout and precise response to prompts.
At the heart of Stable Diffusion 3 lies a novel Multi-modal Diffusion Transformer (MMDiT) architecture. This architecture enables the model to handle image and language representations independently, enhancing its precision in understanding and processing text and increasing its tolerance for spelling errors. This improvement is pivotal in boosting the quality of generated images and their correspondence to the input text.
Stability AI researchers claim that Stable Diffusion 3 excels in human preference assessments. The images it generates are not only visually striking but also demonstrate higher accuracy in adhering to user prompts. This development foreshadows a broader application scope for text-to-image technology in content creation, design assistance, and visual communication.
Stability AI’s innovation is poised to redefine industry standards for text-to-image generation and pave new ways for the integration of AI with art and design. As technology advances, we can anticipate more exquisite images created by AI, concurrently opening up endless possibilities for news reporting, advertising design, and the creative industry.
【来源】https://stability.ai/news/stable-diffusion-3-research-paper
Views: 1