StabilityAI Releases Groundbreaking StableDiffusion3Medium

<p>TapTechNews June 13th news, StabilityAI released StableDiffusion3Medium (hereinafter referred to as SD3Medium), which the official claims is "<strong>the most advanced open source model so far</strong>", and its performance even exceeds that of Midjourney6.</p><p style="text-align: center;"><img src="https://img.taptechnews.com/news/2024/06/1718251371682651.jpg" /></p><p>StabilityAI company said that SD3Medium can generate realistic images according to the text description input by users, focusing on overcoming the challenges of hands and faces in the text-to-image model.</p><p>SD3Medium also utilizes its underlying DiffusionTransformer architecture to precisely integrate text elements with high accuracy.</p><p>Another feature of SD3Medium is its ease of use. Compared to some resource-intensive AI models, <strong>SD3Medium can run on consumer-grade graphics cards</strong>, which can accelerate the popularization and adaptation.</p><p>StabilityAI provides SD3Medium under a non-commercial license for free use. For commercial applications, a creator's license can be provided for artists, designers and developers; for large commercial users, you can directly contact StabilityAI to learn about the licensing details.</p><p>StabilityAI also stated that it plans to expand its products to the field of video and audio generation in the future. The prompt words are as follows:</p><p>A photograph of an 18-year-old Japanese woman hitchhiking, holding a cardboard sign that reads '東京駅まで' (To Tokyo Station). She is standing by the roadside with a hopeful expression, wearing casual clothing and a backpack. The background shows a bustling urban street with cars passing by and city buildings. The scene is lively and vibrant, capturing the energy of Tokyo. Cinematic composition, trending on artstation.</p><p style="text-align: center;"><img src="https://img.taptechnews.com/news/2024/06/1718251373817745.jpg" /></p><p style="text-align: left;">TapTechNews attached the related pictures generated as follows:</p><p style="text-align: center;"><img src="https://img.taptechnews.com/news/2024/06/1718251375757704.jpg" /></p><p style="text-align: center;"><img src="https://img.taptechnews.com/news/2024/06/1718251377140256.jpg" /></p><p style="text-align: center;"><img src="https://img.taptechnews.com/news/2024/06/1718251379207535.jpg" /></p>
Likes