Sign In

UnderPlayer

29
292
1
Verified:
SafeTensor
Type
LoRA
Stats
292
Reviews
Published
Aug 9, 2023
Base Model
SD 1.5
Training
Epochs: 14
Usage Tips
Clip Skip: 2
Trigger Words
Player
grey skin
Hash
AutoV2
283A9A6A49
default creator card background decoration
fuzzy828's Avatar
fuzzy828

Hi guys. This is my first time posting a LoRA on Civitai. I'm not that good at making a LoRA either so I apologize if this model does not generate top quality images. I really liked Undertale and this AU in general so that is why I made this. This model is supposed to generate images of Player from the UnderPlayer AU made by Yunan (鱼腩).

Limitations

This LoRA model is not perfect and it has some limitations that I’m aware of. Some of them are:

  • The model sometimes generates a white part of the dress/sweater at the bottom part instead of remembering that the white is only for the inside.

  • The model sometimes generates extra or fewer stripes on the dress/sweater.

  • The model sometimes generates a regular human skin tone instead of a gray-ish one.

  • The model sometimes generates multiple views of Player in one image.

  • Sometimes like to combine uncolored and colored sclera which does not end well.

  • This LoRA generates colored (black) scleras better than regular white ones.

Settings:

To generate images using my LoRA model, I recommend you use the following settings:

  • Use a 2D style checkpoint instead of a semi-realistic or realistic one.

  • Use a higher CFG scale like 8.5+.

  • For the prompt, along with the trigger words, I recommend you use: red hair, short hair, solo, pantyhose, black boots, (red striped dress), (striped sweater), turtleneck. And then whatever else you want.

  • For the negative prompt, I recommend you use: text, title, logo, signature, watermark, multiple views, white dress, cropped, out of frame, arms behind back, jpeg artifacts, inaccurate eyes, bad hands. And also things like AuroraNegative.


When generating the preview images I used ColorMagination and I recommend you use it if you want to try this LoRa since it generates good images. To make this LoRA I used this resource. I hope you like this and please tell me any feedback or suggestions you might have so I can improve. Thank you for reading all of this.