V3.0:More soap and elbow grease on the dataset. Also isolated her blue dress. She does have several outfits but if i don't remember them they were not iconic enough. Also made her face markings as part of her trigger so it might be hard to get rid of them.Works fine with: &lt;lora:BelldandyV3:0.7&gt;, belldandy,Oufits:Belldandy_Blue_DressAs always I'll add the tag summary inside the dataset so people can experiment.V2.0:Retrained the LORA. As it was my first LORA, The dataset wasn't as good as I would have liked(I have since gained a lot of experience cleaning datasets). Still some crappy images evaded even this second filter.Had nice results with: Belldandy, blue_eyes, brown_hair, (facial_mark:1.3), &lt;lora:Belldandy:.45&gt;Here are some useful tags from the tags summary:<ul><li>long_hair</li><li>facial_mark</li><li>forehead_mark</li><li>blue_eyes</li><li>jewelry</li><li>brown_hair</li><li>smile</li><li>large_breasts</li><li>earrings</li><li>choker</li><li>bracelet</li><li>antenna_hair</li><li>blush</li><li>ponytail</li><li>sweat</li><li>huge_breasts</li><li>bangle</li></ul>V1.0:Her clothes seems to have a strong bias to blue and white.Got good results with &lt;lora:Bdandy:.4&gt;, (Belldandy), (facial_mark:1.1),

Belldandy - Oh My Goddess!

BelldandyV3.safetensors

data.zip

v3.0

v2.0

v1.0

One classic quality Waifu. This was trained with the same dataset as my <a rel="ugc" href="https://civarchive.com/models/8929?modelVersionId=119368">BelldandyV3 LoRA</a>, so download both and compare. I would have hopped for a thicker Belldandy but it is what it is.Works fine with: &lt;lora:BelldandyV3:0.7&gt;, belldandy,Oufits:Belldandy_Blue_DressAs always I'll add the tag summary inside the dataset so people can experiment.

Files with SHA256: 37e5da651eeb7814b305237c9b4633da373943205c71a5afcdbef0f1b56bf334