LoRA trained with 25 images generated using the SD1.5 LoRA https://civitai.com/models/213529/ghost-ogre-and-snow-rabbit-yu-gi-oh-duel-monsters-commission
The training focused on her; her rabbits and weapons were underrepresented. They may still appear, but the LoRA is not designed to generate them.
Check the release notes for more details.
More example images are available on Pixiv https://www.pixiv.net/en/artworks/125236647 (if needed, access the metadata by downloading the image).
Trigger: yuki_usagi
Body description: flat_chest, long_hair, red_eyes, twintails, white_hair, horns, pointy_ears, albino
Clothing: japanese_clothes, black_kimono, obi, sash
Strength: 1 seems to bring out learning biases; I use 0.8
描述:
The kimono tends to appear "misaligned."
The horns have inconsistent sizes.
Rarely, an incorrect number of horns is generated.
训练词语: yuki_usagi
名称: 1167268_training_data.zip
大小 (KB): 1221
类型: Training Data
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
名称: Yuki-Usagi_SDXL.safetensors
大小 (KB): 223099
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
