Ousama Ranking 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 ☕
The ranking of kings. Oddly fitting that this is the first thing I get to publish since getting to #1 on Civitai. Thank you all for the great support during the last few weeks.
Anyway, this triggers with ousamaranking style and, despite being a style LoRA, can reproduce characters that are very similar to the ones of the anime. However, I was still able to make other characters with it using my other LoRA's. This also excels at making scenery in the style of this anime, that I suggest you check out if you never did.
As for the weight you should go between 0.5 and 0.7. I used Abyss Orange Mix 2 for the examples, but AnyLoRA will work fine too.
The first pic is inpainted from pieces of the last one.
Please enjoy and keep supporting me.
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 (by default it's
:1which is usually too high)
Description
FAQ
Comments (13)
looks nice, like some as old cartoon, thanks
It's from this anime: https://www.youtube.com/watch?v=dWZAH5w8jkQ
I suggest you check it out :)
amazing work as always! can i ask do you usually manually caption your datasets ? and how big are the datasets normally, for example how big was this one ?
this one was 924 files. The number is not really important, what matters is the variety. I'd say this could have been improved a bit with some more female examples. I had success with datasets of 20 pics and datasets of 4000 pics.
@Lykon since the number is so high i guess you're doing automatic captioning ? if so how does the LORA recognise the characters ? if you ever get the time a tutorial on your process would be extremely valuable for the community, great job either way !
Thank you for uploading this :)
we love this
maybe I will be back after I try it
Did you use High res ?
If yes what's the parameter please (I don't manage to make similare image)
R-ERSGAN Anime Video at 0.3 denoise and 10 steps. But it shouldn't be necessary, just an improvement
您好，在civitai上看见了您，感觉好厉害。我的邮箱[email protected],方便了可以建联探讨一下 AIGC的落地场景
Sorry I don't speak Chinese. I translated a bit with Google but I'm not sure I really get what you're talking about :)
@Lykon can I leave a wechat message?
@dtt121191500 I don't have wechat
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.