Sign In

Sumi (Monobeno)

98
538
4
Verified:
SafeTensor
Type
LoRA
Stats
538
0
Reviews
Published
Apr 11, 2023
Base Model
SD 1.5
Hash
AutoV2
6F932420A8

This is a LoRA for the Monobeno visual novel character Sumi.

It's a pretty strong one, so I recommend using it with a weight of 0.5 ~ 0.7

This LoRA was trained on the AbyssOrangeMix3 model, but I've found that it works well with many anime models.

There are no trigger words for this LoRA, but to get Sumi's signature look use the following tags in your positive prompt:

red eyes, short hair, two-tone hair, jitome, flat chest, blunt bangs, sumi \(monobeno\)

You can get her signature red kimono by simply adding the red kimono tag as well, but she can be generated in just about any attire or pose you can think up. However, thanks to the training set, if you want your results to be at any other time than day time you should add extra emphasis to your time tags, i.e. (sunset:1.4)

NOTE: Your results may vary from mine if you use xformers due to it's tendency to produce non-deterministic results OR if you use a GPU with a different architecture than mine due to the way a GPU's architecture influences results for various different reasons.

I use several negative embeddings in my negative prompts. They are as follows:
EasyNegative: https://civitai.com/models/7808/easynegative

bad-hands-5: https://huggingface.co/yesyeahvh/bad-hands-5/tree/main

bad-artist-anime: https://huggingface.co/nick-x-hacker/bad-artist/tree/main

Special thanks to WarriorMama777 for putting together AbyssOrangeMix3
Source: https://huggingface.co/WarriorMama777/OrangeMixs