The English description is translated below the Japanese one at DeepL.

重度の厨二病患者である四国めたんのLoRAです。

トリガーワードは　shikoku metan, voicevox　です。

先に作ったずんだもんと比較すると正面からの絵以外が無いので動きが鈍いです。それならと重みを下げると再現性薄くなるので blue eyes, pink hair, long sleeves, long hair, blush stickers, white dress, twintails, puffy sleeves, frills, twin drills, maid headdress,heart hair ornament を入れると服の細かな形状以外は安定します。

重みはAnimagine版は0.5前後、Pony版は0.6前後推奨。

Animagine版はAnimagine XL 3.1をベースとして学習していますが3.0でも動作は確認しています。それ以外のXLモデルでは正常に機能しない可能性があります。

Pony版はPonyDiffusionV6XLをベースとして学習しているのでそれ以外のPony派生モデルでは正常に機能しない可能性があります。

以下翻訳

This is LoRA of Shikoku Metan, a severely kitchen sick person.

Trigger words : shikoku metan, voicevox

Compared to Zundamon, which I made earlier, it has no other pictures than the front view, so it moves slower. If we lower the weights, the reproducibility will be less, so we will use blue eyes, pink hair, long sleeves, long hair, blush stickers, white dress, twintails, puffy sleeves, frills, twin drills, maid, and a heart hair ornament. headdress,heart hair ornament will be stable except for the detailed shape of the clothes.

We recommend a weight of around 0.5 for the Animagine version and 0.6 for the Pony version.

The Animagine version is based on Animagine XL 3.1, but we have confirmed that it works with 3.0 as well. Other XL models may not function properly.

Pony version is trained based on PonyDiffusionV6XL and may not work properly with other Pony derived models.