Shanghai Step Star Intelligent Technology Co., Ltd. has released a major upgrade of its Step-1X series of image generation models - the Step-1X-Medium version. This version has significant improvements in terms of generation speed, understanding ability, consistency of pictures and text, and texture of picture details. It also has a new "picture-generating picture" function that supports detail enhancement, style migration, and local modification. Step-1X-Medium also strengthens its "Chinese style" content creation capabilities and supports adding English copy to Prompt, further expanding its application scenarios and creative space. This upgrade aims to provide creators with a more accurate and perfect image generation experience.
Shanghai Step Star Intelligent Technology Co., Ltd. recently announced that its image generation model Step-1X series has received a major upgrade and launched a Step-1X-Medium version with better performance. This upgraded version has achieved significant improvements in many aspects: based on the MMDit architecture, the generation speed has been increased by more than 30%; after targeted training, the new version has stronger understanding capabilities and consistency of graphics and text, and the generated pictures have detailed texture. More natural.
Step-1X-Medium has added a new "Picture to Make Picture" function. Users only need to upload a picture and give simple instructions to enhance the details, transfer the style or partially modify the original picture. In addition, the new version has also been upgraded in its "Chinese style" content creation capabilities, which can better capture the charm of oriental faces and present a more advanced and delicate picture texture. At the same time, Step-1X-Medium also supports adding English to Prompt, so that the generated image can present English copy.
The new version of Step-1X-Medium has reached a new level in terms of generation quality, can generate more diverse scenes, and has stronger image and text consistency. It can also deeply optimize the image of oriental characters, easily "capture" Chinese style, and generate comic pages with consistent styles for fans of Chinese comics, Japanese comics, and American comics. For brand designers, Step-1X-Medium can generate advertising, product packaging and marketing materials that are in line with the brand's tone and better demonstrate the brand's cultural core.
The "mat picture" function launched by Step-1X-Medium allows creators to upload basic pictures, and the model can quickly understand the picture structure and style, and perform detail enhancement, style transformation or partial refinement based on the creativity of the original picture. In addition, Step-1X-Medium also supports SRef (Style Reference) generation function, providing style reference pictures, and the model extracts aesthetic style and atmosphere features and integrates them into the generated picture composition.
The advancement of AI technology enables Step-1X-Medium to add short text English copy to the prompt to embellish the picture work. This upgrade not only improves the quality and efficiency of image generation, but also provides more creative space and possibilities for creators.
Experience address: https://platform.stepfun.com/
With its powerful functionality and ease of use, Step-1X-Medium will provide artists, designers and content creators with unprecedented creative convenience. It not only improves the efficiency and quality of image generation, but more importantly expands the boundaries of creativity and brings new possibilities to various industries. Welcome to visit the experience address and experience the powerful functions of Step-1X-Medium for yourself!