Type | |
Stats | 29,621 137,286 |
Reviews | (3,399) |
Published | Apr 19, 2023 |
Base Model | |
Trigger Words | nami \(one piece\) |
Hash | AutoV2 3D98234D54 |
Nami (One Piece) LoRA (pre and post timeskip)
CHECK BOTH VERSIONS
Pre Timeskip: https://civitai.com/models/15431?modelVersionId=18188
Post Timeskip: https://civitai.com/models/15431?modelVersionId=18189
Also check your filters to see all the examples
Making models can be expensive. Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕
Well, this wasn't easy to get right. One version of Nami alone is already pretty hard to make: she has a very specific tattoo, asymmetrical, she has exactly 2 rings per side on her jeans (or on her skirt before timeskip) and she wears tons and tons of different official outfits. Plus her artworks tend to be inconsistent, so I had to clean the dataset a lot. Not to mention the timeskip problem: she is basically 2 different characters, both very hard.
At the end of the day I had to accept some compromise. First of all I had to make TWO LoRA's and not one. Second I needed to focus on getting her main outfit right and, most importantly, the tattoo. Everything else came second.
I think I managed somehow. I made 2 LoRA's, one for pre timeskip Nami and one for post timeskip Nami, both of them can do both versions, but they focus on their respective ones.
Main outfits are always kind of right, but exactly right about 1/4th of the time. Tattoo is almost always right. Secondary outfits are almost right about 1/4th of the time, depending on how rare they are in artworks.
The model is not overfitted and can be used on most booru based models. I mostly tried it with AnyLoRA and NeverEnding Dream for realistic shots. Weight should be around 1 for post timeskip and around 1.2 for pre timeskip for better accuracy. I suggest using weight 1 if you need just her face and tattoo on non-official outfits.
Triggers with nami \(one piece\)
for better accuracy. Also use short hair
in pre timeskip version. One of the examples is also using my Wano Style LoRA.
Please enjoy.
How to use LoRA's in auto1111:
Update webui (use
git pull
like here or redownload it)Copy the file to
stable-diffusion-webui/models/lora
Select your LoRA like in this video
Make sure to change the weight according to the instructions (by default it's
:1
)