Conversation
|
PRありがとうございます。素晴らしいですね。確認させていただきます。 |
|
when exactly would you want to use something like this? Im not clear on the effect that this has |
see here |
Negative weightを使ってもいいですか? |
Of course you can use negative weight, but i'm not sure how it would affect. |
|
|
|
なるほど、分かりました。 |
|
@u-haru Thanks for your work! |
|
Hello, @ymzlygw ! In terms of training data, I use approximately 10 to 50 images with several angles and white background for both training. I typically train with batch_size=2 for around 2000 to 3000 steps, but I think it would be better to train more steps on first training. |
Thanks! I'll try it later! |


学習開始前に学習済みのLoRAモデルを適用して、差分のみを学習するオプションを追加します。
複数のモデルを適用できるようにはなっていますが、2つ以上適用した場合の効果は未検証です。
差分学習の効果についてはエマノンさんの記事(note.com/emanon_14/n/ne83063e33627)をご覧ください。
以下は私の検証です。適当に3Dのモデルを学習させました。
1枚目はそのまま学習、2枚目は3Dの画風を学習したモデルをweight=0.9で適用した上で学習したものです。
2枚目の画像は3Dっぽさがなくなっていると思われます。