h2iai Fullshot BeautySaga

The purpose of the LORA training is mainly to improve the quality of generating full-body female characters. It has been tested and can be effectively combined with various large models, including 3D, 2D, and 2.5D. By using the same parameters and corresponding base models, similar results can be reproduced.

The prompt is straightforward and does not require additional images or control nets.

The large models used for generating the sample images, from left to right, are: moonfilm fantasy, majixmixRealistic, ghostmix, fantasticmixreal, dalceforealistic, dalcefooilpainting, AWportrait, cyberrealistic, chilloutmix, AWpainting.

To avoid facial distortions, you can use ADetailer or face editor instead of facial restoration. Personally, I recommend ADdetailer, which can be downloaded from webui extension tab.

I will continue to use this LORA for testing and upload the results for your reference.

该LORA训练的目的主要用于提高生成全身像女性角色的质量，经测试可以较好的和各类大模型混合使用，3D/2D/2.5D均可，样图使用相同的参数和对应的底模基本可以复现。

Prompt直出，无需垫图，无需controlnet。

样图使用的大模型从左往右以此为：moonfilm fantasy, majixmixRealistic, ghostmix, fantasticmixreal, dalceforealistic, dalcefooilpainting, AWportrait, cyberrealistic, chilloutmix, AWpainting.

需要使用ADetailer, 或者 face editor 解决脸部崩坏问题，不要使用脸部修复，个人推荐ADdetailer, 可以在extension下载使用。

我会持续使用该LORA测试出图并上传已供参考。