找回密码
 立即注册
查看: 1500|回复: 0

[传统艺术] [SD 1.5][Checkpoint]Uhmami v1.0

[复制链接]

3869

主题

36

回帖

4331

积分

管理员

积分
4331
发表于 2023-6-7 21:40:31 | 显示全部楼层 |阅读模式

基础数据

样图
基本模型:
SD 1.5
小模型:
Checkpoint 
下载链接:
您必须是网站正式用户方可进行下载。

Uhmami v1.0

This model is a blend of anime-focused models with a semi-realistic aesthetic, combining 3D digital imagery with photography. It allows the use of prompts based on booru tags, similar to the novelai model, resulting in high-quality semi-realistic images.

The model excels at creating images of young women and can even handle NSFW prompts, although it may not be the best at representing explicit sexual content. However, it's just as effective at creating images of men, animals, and landscapes, as seen in the sample images.

One limitation to note is that it may be somewhat limited in its ability to provide female faces for similar prompts. However, in some cases, this can be an advantage in maintaining visual coherence across multiple images.

Usage recommendations:

  • We highly recommend using the negative prompt (low quality:1.4) as it has a tremendous impact on improving the resulting images. You can combine it with other negative prompts to further enhance the result, but this one is essential. Without it, the resulting image may be flat and of poor quality.

  • Because it incorporates anime models, it tends to represent Asian features, but this can be easily avoided by using the appropriate negative prompts.

  • While not necessary, we recommend using VAE, specifically the vae-ft-mse-840000-ema-pruned model, for best results.

  • Additionally, using Clip skip 2 in the configuration can add depth and richness to the resulting images.

I hope you find it both useful and enjoyable to use!

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

Archiver|手机版|小黑屋|金房子

GMT+8, 2024-12-23 20:08 , Processed in 0.040288 second(s), 23 queries .

© 2023 金房子|AI发烧友社区

快速回复 返回顶部 返回列表