打率は低いが沙村広明っぽい線は出てくる
PaperspaceでSDXL学習が出来るようになったのでテストで作ったが演算コストがバカ高(設定値探るの含めて2.5万円)でわざわざA100を時間貸しで作成したから次のバージョンはない
"The batting average is low, but there are lines reminiscent of Hiroaki Samura, the author of 'Blade of the Immortal' (無限の住人).
I was able to train SDXL on Paperspace, but the computational cost was ridiculously high since I had to rent the A100 for hours, so there won't be a next version."
SD1.5で作ったLoraは無限の住人､ベアゲルター､ひとでなしの恋のおかげでSMだとかグロ､リョナに強かったが､これはベースモデルの語彙の問題でその手の表現がマイルド
v1.5
次のバージョンはないといってなあれは嘘だ
Ponyベースに変更
完成度上がった
Description
1.5にベアゲルターLoRAの画像を追加
FAQ
Comments (2)
Why did you need an A100 to train? It's possible to do SDXL LoRA with 24GB or even 16.
Six months ago, there wasn't a lot of understanding of SDXL or research on its parameters.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.