[Model]ラトナ・プティ/ratna_petit (Patreon)

*本作品に登場する人物はすべて成人済みです

//////////////////////////////////////////

*All characters in this work are of legal age.

*何件か要望をいただいてるのでそろそろぶいすぽもやります(>_<)

(他Vtuberも今まで通り投稿続けます)

////////////////////////////////////////////////////////////////

*I've had a few requests for this, so I'm going to start doing vspo! soon.

(Other Vtubers will continue to post as before)

*他支援サイトの更新の関係で来月一回投稿整理したいと思います!

実はかなり古い投稿も混ざってるので・・・

そのため過去投稿のDLは今のうちに行っていただけると幸いです(>_<)

Patreonでは今のところ問題なさそうなので今後も投稿続けます!

ご支援いただき本当にありがとうございますm(__)m

////////////////////////////////////////////////////////////////

Due to updates on other support sites, I'd like to organize my posts once next month!

Actually, there are some very old posts mixed in...

So we would appreciate it if you could do a DL of your old posts while you are at it (>_<).

Patreon seems to be working fine so far, so I will continue to post!

Thank you very much for your support. m(__)m

オリジナル画像(stable diffusionのPNG Infoタブでご利用ください)

////////////////////////////////////////////////////////////////

Original image (please use the PNG Info tab in stable diffusion)

*LoHaモデルです。

000018~000020推奨です。

LoHaとはloraよりも出力に優れた最新のモデルになります。

最新のStableDiffusionであれば、loraと同じ使い方で問題ありません。

念のため実装方法は下記に追記してますのでご確認ください。

また、学習率ごとにモデルを載せています。

数字が大きいほど学習率が高く再現性も上がる傾向にありますが、過学習気味になったり、似たような構図が増える可能性があります。

////////////////////////////////////////////////////////////////

LoHa Model.

000018~000020 recommended.

LoHa is the latest model with better output than lora.

If you use the latest StableDiffusion, you can use it in the same way as lora.

Please refer to the following for implementation instructions, just in case.

Also, the models are listed by learning rate.

The higher the number, the higher the learning rate and the higher the reproducibility, but there is a possibility of over-learning or more similar compositions.

parameters:

1girl,solo, highres, <lyco:ratna_petit-000018:0.6>,ratna_petit,orange_eyes,orange_hair,short_hair,ahoge,animal_ear_fluff,black_jacket,jacket_over_hoodie,open_clothes,open_hoodie,open_jacket,orange_hoodie,red_bowtie,plaid skirt,

panties ,(lifted by self:1.4),from above,naughty face,looking up,canal,closed mouth,jumping,

Negative prompt: (worst quality, low quality:1.4),

Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 7, Seed: 3927937568, Size: 512x768, Model hash: 5493a0ec49, Denoising strength: 0.45, Clip skip: 2, ENSD: 31337, Hires upscale: 2, Hires steps: 13, Hires upscaler: 4x-UltraSharp

*現在は多様な設定項目があり、出力を完全に一致させることは難しいです。

概ね出力が一致していれば多少の差異があっても正しく適用されているためご安心ください。

////////////////////////////////////////////////////////////////

Currently, there are a variety of configuration items and it is difficult to match the output exactly.

Please be assured that if the output is generally consistent, even if there are some differences, they are correctly applied.

実装は下記投稿を参考にしてください。

////////////////////////////////////////////////////////////////

Please refer to the following post for implementation.

実装方法リンク/Implementation Method Link:

https://www.patreon.com/posts/moderushi-zhuang-81561102