The model is suitable for generating both authentic and anime images, including 3D, 2.5D, and 2D Figs. Real but unreal, Just like the Utopia that only exists in the idealistic world, that is the meaning of word 'virtual'.

The main image includes people, animals, plants, extraterrestrial organisms, buildings, etc., all of which can obtain high-quality images.The prompts should be as concise as possible to achieve better results.

The art style is diverse, including mainstream styles such as Impressionism and Surrealism, and niche styles are also included. Using grid based plugins: ‘ https://github.com/ahgsql/StyleSelectorXL.git ’ can help you remember various style names. (Me cyber cat thinks this plugin has the same function as using prompt words, it just reminds you that there are so many art style prompts.)

The initial model is not embedded SDXL vae, but later cat me came up with a unique way to embed sdxl vae into the model, therefore, using or without vae during generate can result in nice quality images.

Suggested Ksampler:

WebUI: Euler a \ DPM++2M Karras \ 2M SDE \ Restart

ComfyUI: DPM++3M SDE Karras \ DPM++3M SDE Expert \ 2M \ Restart

Also, You can try other samplers and multiple samplers on your own to archive a better effect.

模型适用于真实风格与动漫风格，包括3D、2.5D及2D图片，就像只存在于理想主义的世界里的乌托邦一样——真实且虚幻。

主体形象则包括人、动物、植物、外星生物、建筑物等，都可以得到质量较高的图片，在绘画时提示词尽量精简，效果会更好。

画风多样，不仅包括印象派，超现实主义等等主流风格，小众风格也内含其中，使用“sdxl风格插件https://gitclone.com/github.com/ahgsql/StyleSelectorXL ”可以帮助你记忆各类风格名称（猫子觉得此插件和用提示词效果一样，插件只是提醒了你有如此之多的画风prompts）。

此初版模型不内嵌SDXL vae，后来猫子我又想到一个奇特的办法将vae融入模型中，在使用时调用或不调用vae皆可。

建议采样器：

webUI: Euler a\DPM++2M Karras\2M SDE\Restart...

comfyUI: DPM++ 3M SDE Karras\DPM++ 3M SDE Exponential\2M\Restart\....

可以自行尝试其他采样器及多个采样器，会有惊艳的效果。

DPM++ 3M SDE Karras

Restart

DPM++ 2M Karras

DPM++ 3M SDE Exponential

Euler a