About SD Style Lab

Note: The translations on this site are generated by LLMs. If you notice any errors or inaccuracies, please refer to the original Chinese text as the authoritative version. Thank you for your understanding.

SD Style Lab is a site dedicated to organizing and displaying AI image style comparison records, with a focus on the actual generation effects of different "artist tag combinations" under various models.

What This Site Does

Today's anime image generation models, whether it's the early but closed-source NovelAI or the later outstanding open-source models (such as NoobAI XL), already possess strong generation capabilities. With the right prompt combinations, they can produce images with distinct styles and rich expressions.

However, these models also have a notable characteristic: if you want to consistently trigger certain high-quality art styles, specific "artist tag combinations" often need to be added to the prompts, such as konbu wakame,xilmo,wlop,year 2025,. These combinations can indeed significantly improve the artistic style of the generated images, but they also raise the barrier to entry.

Unlike relatively intuitive common prompts like 1girl, walking, or raining, artist tag combinations are not easy to organize, and it's difficult to predict the final result from text alone. In many cases, you only know what style it will produce under the current model after actually running the generation.

Although there are already some existing artist tag compilations and test results on the market, such as the "300 Art Style Codex," they are usually distributed in xlsx or online shared documents and often only test a single model. This brings two problems: first, such materials mostly circulate in small circles, making them difficult for newcomers to find and get started with; second, performance varies greatly between different models, so single-model test results are hard to cover actual usage needs.

What SD Style Lab aims to do is organize these scattered, fragmented, and difficult-to-compare comparison records into a website that is easier to browse and search, continuously displaying the generation results of different artist tag combinations under different models, and constantly adding new model test data to help users more efficiently understand and use these powerful image generation models.

What You Can See Here

On SD Style Lab, you can browse the generation results of different artist tag combinations under different models and intuitively compare the differences between them. As test data continues to be added, you can also more easily observe the performance of the same set of artist tag combinations under different models, allowing you to choose more suitable models and prompt combinations based on your creative needs.

About Login and Access Permissions

To prevent excessive use of the site, you currently need to log in to view the full content. The login process is simple, supporting GitHub, Google, and Microsoft accounts. After logging in, you can access all content on the site (including NSFW content).

About Data and Image Assets

This site strives to provide reliable data and image assets. The images displayed on the site are either generated by this site itself or sourced from publicly available resources.

Current Stage

SD Style Lab is currently in the testing phase, and both the site's features and content are continuously being improved. If you have any questions, suggestions, or feedback during use, feel free to submit an issue on the GitHub repository. Your feedback will help me continue to improve the content organization and user experience of this site.