Ultimate Hentai Wildcards

详情

下载文件 (1)

模型描述

"VITAL" INFO:
I made this right after learning what wildcards are so they may be a bit wonky. In my testing I have not found any major problems. I do not know how well this works with a LoRA since I have not tested it, but I cannot imagine it being incompatible.

If there is anything you dislike in the wildcards fix it yourself. If you think it is a change others would appreciate leave a comment explaining and I will implement it if I see fit.

The clothes do not have that much reliability just like without the wildcards. It may just be because I am bad at prompting.

PROMPTS:

Recommended negative prompts:

broken limbs, torn clothes, damaged clothing, broken zipper, three arms, four arms, double forearm, fucked up generation, three boobs, twisted knee, three legs, ass on wrong side, bad eyes, nudity,

I also recommend the negative embeddings I use. You can find the links at the bottom of the description. The settings I use for the embeddings are:

(By bad artist -neg, easynegative, FastNegativeV2, (negative_hand-neg:1.2), ng_deepnegative_v1_75t:1.5),

They work like a charm!

Most of the problems that the recommended prompts are to fix are (from my experience) mostly appearing when creating a wallpaper. I only added "fucked up generation" since I was tired and it worked so I cannot be bothered to change it.

Do not forget to put double underscores before and after the text file name! Example: __UHboobsandass__

PATREON:
I have made a Patreon! You can and will get early access to my models there. It costs 5$ to subscribe. I am not expecting anyone to actually subscribe but I would appreciate it immensely!

RANDOM INFO:

The wildcards in my prompts do not have UH because I want to make it easy for myself. The only reason why I added UH was so that it hopefully would not clash with existing wildcards that you may use.

UH is short for Ultimate Hentai

LINKS:

Deep Negative V1.x

Bad artist Negative embedding

EasyNegative

negative_hand Negative Embedding

Fast Negative Embedding (+ FastNegativeV2)

此模型生成的图像