A character LoRa for Nami from One Piece. This was trained with screencaps from episode 1 of the anime series using part of the dataset of my 10 outfit LoRa for her. It's trained to generate the two different outfits she is wearing in episode 1. The dataset for the dress (ball gown) is very limited, so that outfit is not at the level of accuracy I want it. I was able to improve the image quality for that outfit in ill_v2.0 (and anima) version though, in ill_v1.0 it looked kinda blurry.
pirate outfit trigger words: nami, 1girl, prt, orange hair, short hair, brown eyes, bandana, blue shirt, wristband, red sash, black pants
dress trigger words: nami, 1girl, psdr, orange hair, short hair, brown eyes, dress, short sleeves, earrings, red choker, bracelet, red high heels
Feel free to comment, if you have something to say about the lora and share your images on the model page. Sharing your images on the model page gives the creator a small blue buzz bonus. But more important I like to see, what people do with my LoRas.
If you like the LoRa please have a look at the showcase pics and leave a reaction.
Description
New version trained with slightly improved dataset and tagging. Unfortunately there isn't any artwork of her ball gown from episode 1 that I could find, so I had to work with the few anime screenshots I had. I retouched them a bit to reduce the blurriness and it did help along with using different tags for the style. However since there was only one image of her lower body (and only of her lower body) wearing the dress the details of the pattern did not bake in (there should be more lines than the one at the hem). If anyone knows about accurate artwork of this outfit, please let me know (If anyone who can draw fairly good is up for trying to draw the dress and share it, that would be awesome). I used tags "anime screenshot", "anime coloring" and "opscreencap" for the style, so you can put that in the negative prompt if you want to reduce the style influence. Unfortunately there is a bit of concept bleeding between the two outfits (I guess due to the different captioning), so put the outfit trigger word of the other outfit in the negative prompt to avoid that (also clothing pieces of the other outfit like "choker" or "sash").