该模型使用了17723张图片训练，覆盖了大部分EVA主要角色。
在4090上训练了50小时。
请注意，对于pony模型来说， EVA大部分角色属于已有概念。因此该模型更侧重于对画风的改善（更加日漫风格）和概念补全。
This model was trained using 17,723 images, covering most of the major EVA characters.
It was trained for 50 hours on 4090.
Please note that for the pony model, most of the EVA characters fall under the existing concepts. Therefore, this model focuses more on improving the art style (more anime-style) and concept completion.
Training resolution: 1024*1024
Recommended output resolution:832*1216/ 1024*1280
Training model: pony
Inference model: AutismMix SDXL
特别鸣谢：算力赞助 nieta.art / GPU Sponsor: Neta
Recommended dataset collection and processing tools: https://github.com/deepghs/waifuc
Possibly the most user-friendly dataset processing UI currently in use： https://github.com/Eugeoter/sd-dataset-manager
Description
FAQ
Comments (10)
This is great! If you ever decide to make a V2 it would be fantastic to see some of the non-main cast members like Hikari Horaki, Kensuke Aida and Toji Suzuhara.
I will consider adding other characters. The only concern is how to gather the images of these characters.
@Chenkin Have you considered using movie screencaps as data as well? Or maybe have specific trigger words for different plugsuits?
would you be willing to add some kind of prompting to the description of this lora like you did for the touhou and genshin character loras?
Is there any way we could get Kaji and Ritsuko added to this?
I know this is a super late response, but I have a Kaji model I baked, and Ritsuko is definitely baked into this LoRA.
Lmao, you added Makinami which isn't even existed in original Eva, but no Ritsuko?
Ritsuko is baked into this LoRA, i use it all the time when I want to generate her
Most examples are "just pilots", which ain't too bad
Do you have trigger words for not just characters, but specific outfits (e.g. plugsuit) and scenes/objects (e.g. entry plug cockpit, EVA units, stasis pod)?
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.