Studio Ghibli style LoRA
Making models can be expensive. Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕
I know there are already various Ghibli models, but with LoRA being a thing now it's time to bring this style into 2023. This is good around 1 weight for the offset version and 0.65 for the old one, on Anything v4.5 for a more authentic style, but it's also good on AbyssOrangeMix2. Triggers with ghibli style and, as you can see, it should work with any character LoRA or TI that didn't overfit on style too. Keep in mind, this is a style model not a "Ghibli characters" model, so movie characters in the examples are made using careful prompting, meaning it can reproduce similar characters, but it won't make them perfectly (unless you also use other TI/LoRA's).
4th pic by Karkaz using also his own character LoRA.
Last pic by LottaLewds.
How to use LoRA's in auto1111:
Update webui (use
git pulllike here or redownload it)Copy the file to
stable-diffusion-webui/models/loraSelect your LoRA like in this video
Make sure to change the weight if needed (by default it's
:1)
Description
FAQ
Comments (26)
Looks great.
Can you do Yusuke Murata art style?
I've already done it.
all of mine come out super unsaturated any fix to that
since 3 months ago i have seen some amazing ghibli upscales on social media, this lora could create a nice community, but im happy it didnt, ghibli stuff by midjoureny and nijijourney destroyed the quality tag of ghibli, you know all this miyazaki and ai stuff, It's good that this style model is somewhere in between of serious / humor
Super cool! How hard would it be to build a larger model, possibly checkpoint for specifically 1995 Evangelion series? I searched entire civitai and there is nothing like it.
That's funny. I have wondered about the same thing and I've tried to do something similar. I just don't have the patience for all the tinkering!! 😱
I'm not that attached to this style but this LORA is straight up the best out there. I can plug it in and it just works, it doesn't fight me or deep fry itself at all.
What exactly is an "offset"? Why is there an "offset" version?
offset version can be used at weight 1. The old one is the same but requires you to manually lower the weight.
Could you share some tips for promping with the dataset? What was your prefix and postfix, and the movie name like "kiki" was also added? And also training parameters configurations.. Thanks.
Movie name is not added. There is no particular adivise, use booru style tags. Look at examples, prompts I used are all public
adding stuff like "anime screencap" "film grain" and "1990s style" help get images to look cinematic or a little better in my experience
Hi, would it be a problem to adapt this Lora for SDXL1.0 version?
This model will have stuff related to Wolf Children?
I would assume not, given that Wolf Children was made by a different studio (Chizu as opposed to Ghibli).
I have a collection of pictures of distinctive drawings by artist gaston18
I would like someone to make a professional lora style model for artist gaston18
I have prepared the images in a zip file
Some of these pictures were purchased by me
So this file contains rare and distinctive designs
I was keen to get as many pictures as possible
So if you know someone who can do this
Thank you - a great lora!
I used this LoRA and added the 'ghibli style' trigger word to the prompt, but I didn't get this style. The weight is set to 1. Why is this happening?
Same here.
@calendarmattia389342 any changes?
can anyone share me the SD 1.5 ? download link
¿Where are the memes?
I feel like this lora is going to be intensively used after the openai backlash on social media
I'm here because of the memes