モデルカード / Model Card
モデル名 / Model Name: yoiko_style_model
親モデル / Parent Model: WAI-NSFW-illustrious-SDXL Ver15.0
ライセンス / License: Illustrious License
学習データ / Training Data: 私が生成した画像（未公開を含む） / Images generated by me (including some not publicly released)
紹介文（日本語）
本モデルは WAI-NSFW-illustrious-SDXL Ver15.0 をベースに、追加の SFT（Supervised Fine-Tuning） を施した公開用ベースモデルです。絵柄は過去の yoiko_lora シリーズに近い傾向があります。これは、これまでの学習データと使用画像が類似しているためで、yoiko_lora シリーズが持つ低レベル表現の重みをより丁寧に更新することを目指しました。
ベースとして WAI-NSFW-illustrious-SDXL Ver15.0 を選んだ理由は、圧倒的な完成度と、noob ai 系と比べてライセンスが比較的緩やかで試行錯誤の幅が広い点にあります。現時点でも一定の完成度はあると考えていますが、生成画像の品質には未知数な部分が残っています。ぜひフィードバックをお寄せください。
それでは、楽しんでご利用ください。
Description (English)
This public base model builds upon WAI-NSFW-illustrious-SDXL Ver15.0 with additional Supervised Fine-Tuning (SFT). Its visual style is close to my previous yoiko_lora series. That’s because the training and reference images are similar; the goal here is to refine and update the lower-level representational weights that characterize the yoiko_lora line.
I chose WAI-NSFW-illustrious-SDXL Ver15.0 due to its excellent baseline quality and a more permissive license (compared to some noob-ai variants), which allows broader experimentation. While I believe the model is reasonably polished, output quality may still vary, and there are unknowns to explore. Feedback is very welcome.
In short—have fun creating!
Description
FAQ
Comments (7)
I’ve released Version 1.5 (Experimental)!
Your feedback is very welcome — please feel free to share your thoughts in the comments.
Images generated by v1.5 sometimes have unnatural gloss/lighting, and if the breast size is not specified, the default is always large breasts.
I think it doesn't feel as stable as v1.0.
@Pavers Thank you for the comments!
That’s true, so I’m going to publish another experimental version before V2.0.
The version is already finished, so I’ll publish it in a few days.
You might have already noticed, but V1.5 reacts quite strongly to weights. On the other hand, it doesn’t perform as well when using simple or generic word lists in prompts.
This is probably because I tend to use weights quite often in my own workflow.
Sorry to be late for publish v1.6.
It has the same name of experimental ,but I think v1.6 is better than v1.5.
Please try it!
Also this time I want feedback.
If you have feedback ,please comment.
v1.6 feels slightly better than v1.0 when using DPMPP_2M + sgm_uniform, especially in terms of facial performance. However, the improvement is very marginal; at times, its performance is merely on par with, or even worse than, v1.0.
I'm also starting to get a rough understanding of why hands in the generated images are so prone to errors because I just noticed that this checkpoint is based on wai's checkpoint.
@Pavers I see. Personally, I felt that it was more stable than v1.0, but I realize I shouldn’t rely too much on my own subjective impression.
Indeed, looking again, even the sample images I provided have some instability in the hands.
Since this version is a merged model, forcing the weights together might have introduced some roughness.
That’s why I’m planning to perform SFT again for version 2.0.
I already have an idea of how to approach it, but it will probably take some time.
Thank you so much for your valuable feedback and insights.
I’d really appreciate your thoughts again if there’s another opportunity.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.