`` Advanced Prompt Tuning (APT) '' that dramatically improves the quality and diversity of illustrations with image generation AI `` Stable Diffusion ''
Since the image generation AI 'Stable Diffusion' was released to the public in August 2022,
GitHub - 7eu7d7/DreamArtist-stable-diffusion: stable diffusion webui with advance prompt tuning
https://github.com/7eu7d7/DreamArtist-stable-diffusion
propose an advanced Prompt Tuning method (APT), can super dramatically improve the image quality and diversity by 7eu7d7 · Pull Request #2945 · AUTOMATIC1111/stable-diffusion-webui · GitHub
https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/2945
APT was developed by 7eu7d7 . The example published by 7eu7d7 looks like this. First, the following one image is used as the learning source.
Below is an image generated by letting the conventional Stable Diffusion learn the above image. Even if multiple images are generated, only images with similar compositions can be generated, and some images are broken as illustrations.
And the image generated using APT is below. Compared to images generated by conventional Stable Diffusion, the quality is overwhelmingly higher, and you can see that the composition and poses are disjointed.
In addition, the image generated by adding detailed prompts is below. The hands and feet are drawn in a natural shape, and the quality is so good that you won't notice that it was generated by AI.
7eu7d7 is developing APT as an open source, and at the time of writing the article, development is underway to enable APT to be used as an extension of Stable Diffusion's highly functional UI ' Stable Diffusion web UI (AUTOMATIC1111 version) '. increase.
In addition, 7eu7d7 is also developing an application that collects pixiv images using AI.
GitHub - 7eu7d7/pixiv_AI_crawler: Base high-quality mass map AI reptile,Kaikai your XP
https://github.com/7eu7d7/pixiv_AI_crawler
Related Posts: