hellooo, here’s a lora extracted from my darkvictorian artstyle model! it was one of my first dreambooth models so it’s a bit rough around the edges but i love how the LoRA blends with luna diffusion.
sometimes the A1111 web ui doesn’t like .pt files and I haven’t found something yet to convert them into safetensors. you can use –disable-safe-unpickle in the launch cell if it doesn’t work.
examples / comparisons
- the model uses “darkvictorian artstyle” as trigger words
- has a tendency to put artifacts & weird masks on faces, as seen on the right. can look cool but if you don’t like it you can try using “face mask, blemishes, face paint, glasses” as negative prompt to combat that
- was trained on 512×512 so if you have trouble with proportions in portraits it can help to start with that resolution -> hires fix (as opposed to the 768×768 with luna diffusion)
- biased towards lighter complexions
- seems to not like to depict men lol
hope you’ll like it! ヽ(・∀・)ﾉ