Nagatoro V1

This model aims to help in producing images of Nagatoro, from "Ijiranaide, nagatoro-san". It was trained on 22 images, for 10 epochs, and it is intended to work with pony xl models, in particoular my "Blue-tear" model.

How to use it?

You can pretty much plug and play the model.

To activate it use the keyword: "N4G4T0R0".

Then you can describe the desired attire, and scene.

---

!!Important!!

The model was trained omitting the keyword "1girl" so, when you use this lora, try to not use this keyword cause it's not needed, and might make the generation worse

Example of Positive Prompt

Score_9, score_8_up, score_7_up BREAK <Nagatoro:1> N4G4T0R0, white shirt, upper body, cowboy shot

---

What are the suggested settings?

Lora weight: 0.6 ~ 1

Lowering the weight reduce the impact of the lora on the style, but "some times" require more keywords, for example lowering at 0.7 or below makes nagatoro lose the "dark_skin".

Those are some keywords that can help obtaining the standard nagatoro:

dark_skin, long hair, black hair, brown eyes, hairclip, earclip

---

This model works decently with manga style, just add (at the end of the prompt):

greyscale, comic, screentone

But if you are interested in generating those kind of images, you might wanna get familiar with some additional tag for image composition (ie. koma tags):

https://danbooru.donmai.us/wiki_pages/tag_group%3Aimage_composition

---

Extra tips

The model was trained on 1024*1024 images, so some prompts might struggle with some thinner resolution like 768x1344, but work perfectly fine with wider one like 832*1216.

You can even use the "nagatoro" tag, with "N4G4T0R0" and have an even better generation, but in those cases you should sensibly reduce the lora weight (0.6).

---

Does it support nsfw?

Yes, it does. That said, just because it can do it, it doesn't mean you should do it.

描述:

This is a test version for pony training. It is better than expected, but it could be improved so i might release a v2 in the future.

训练词语: N4G4T0R0