Movie Portrait - V1
ハイライトされた画像
推奨パラメータ
samplers
Euler, DDIM
steps
30
cfg
1 - 7.5
resolution
512x512
ヒント
To get the best results, use the trigger phrase 'In the style of ohwx', though the model performs well without it.
バージョンのハイライト
Trained with "In the style of ohwx" as trigger, but may work without.
Based on portrait images in a cinematic style.
This model is surprisingly good at lending a more natural, photographic look to images without imposing an overly strong effect or messing up prompt adhesion / cohesion.
It does however seem to have a "bug" where on some prompts vertical banding appears.
Trigger words are "In the style of ohwx" but works well without too.