vindictusのキャラクターベル

v1.0

画像枚数72、pixelAIでトレーニング

v2.0

画像枚数80、repeats:2、Epochs:10、batch_size:1、Steps:1580

v1.0とv2.0のトリガーワードが違うので注意

v2.0着せ替えをしやすくしました。

着せ替え時、元衣装が混ざる場合はLoRA重みを減らし、

ネガティブプロンプトにbelle_vindictus_default_outfit,

black dress, race trim, blue skirt, wide sleeves,など

元衣装の要素を追加してください。

Note that the trigger words are different between v1.0 and v2.0.

v2.0 makes it easier to change outfits.

When changing outfits, if the original outfit is mixed in, reduce the LoRA weight and

add elements of the original outfit to the negative prompt, such as belle_vindictus_default_outfit,

black dress, lace trim, blue skirt, wide sleeves, etc.

-

v1.0

trigger words = belle \(vindictus\),

Additional words = light blue hair, short hair, swept bangs, blue eyes, braid, blonde streak, fake horns hat, black and blue dress, skirt, detached sleeves, painted fingernails, socks, boots,

negative = earrings, Yellow hair, shoulders, pouch, black skirt, Black corset, balck underbust

v2.0

trigger words = belle_vindictus,

Additional words = light blue hair, short hair, swept bangs, blue eyes, single braid,

clothing = belle_vindictus_default_outfit, fake horns helmet, black dress, race trim, blue skirt, wide sleeves, painted fingernails, black socks, boots,

negative = earrings, Helmet strap, Yellow hair, Bare shoulders, pouch, black skirt, Black corset, balck underbust, Inner sleeves, Extra tassels, Extra braid