Sign In

Mink (Dragon Half)

124
730
82
7
Updated: Oct 5, 2024
characteranimewoman
Verified:
SafeTensor
Type
LoRA
Stats
449
82
Reviews
Published
May 3, 2023
Base Model
SD 1.5
Training
Epochs: 9
Usage Tips
Clip Skip: 2
Strength: 1
Trigger Words
mink_\(dragon_half\)
bikini armor
dragon tail
dragon wings
dragon horns
knee boots
Hash
AutoV2
64B63F772B
default creator card background decoration
4as's Avatar
4as

Mink is the main character from Dragon Half.

The main trigger is mink_\(dragon_half\), but it also includes: bikini armor, dragon horns, dragon wings, dragon tail, and knee boots, if you want to mix and match various parts. Additional the LoRA comes with retro artstyle trigger to give Mink the original anime look. Weight 0.8 - 1.0 is recommended.

Activation tags

1) mink_\(dragon_half\) is the main trigger. Technically it shouldn't add anything else, but other tags might still bleed into the image. If that turns out to be the case I recommended either decreasing the weight, or putting other activation tags into the negative prompt. If you decide to lower the weight then I recommend adding red hair and red eyes tags too.
2) dragon horns are a bit hit-or-miss. The official art sometimes portraits Mink as have three pairs of horns, while anime and manga show only a single pair. This tag is used to differentiate between them. I actually suggest putting this tag into a negative prompt, as it looks better in my opinion.
3) bikini armor creates the clothes Mink is usually wearing.
4) dragon wings and dragon tail are self explanatory.
5) knee boots are the golden metal boots Mink is wearing along with her outfit.

Bonus tags:
a) retro artstyle forces the anime aesthetics.
b) illustration and official art can be used to shift the art style toward the official illustrations.

Misc.

Be warned, easynegative doesn't gel well with Mink's eyes.

1.0 is obsolete and shouldn't be used, but I'm keeping it as it can produce some interesting results with unique art style.

Trained on 89 images (x5 repeats), with person used as a class. 8 epochs, 64 DIM / 32 Alpha. Base model used for training was animefull_final, but the example images were created on AbyssOrangeMix.

My first attempt at making a LoRA. Feedback is welcome.