タイプ
TextualInversion
投稿日時
2023-03-20
基本モデル
SD 1.5
トリガーワード
emb-jordyn
コピー
バージョン紹介
base version. 1200 steps ate variable learning rate, additional steps at .0005 till 16350



















Embedding trained on images of influencer/actress Jordyn Jones. Works well in RinaMix2 (see samples above). The last 2 samples are image embeddings at lower weight, in case the base version is overtrained for your preferred checkpoint. Base is 16350 steps, samples are 9000 and 13000 steps. They can be dragged into the Embeddings folder just like .pt files.
The training picked up on the fact that she has freckles/beauty marks, but distributes them randomly around her face.
Hope you enjoy. Let me see how it turns out.
base version. 1200 steps ate variable learning rate, additional steps at .0005 till 16350
1.転載モデルは学習・共有目的のみで使用し、著作権は原作者に帰属します
2.モデルの認証は公式チャンネルでご連絡ください。クリエイターの権利保護に努めています クリックして認証
