Bad Quality Lora | SDXL - v2.0
推奨プロンプト
bad quality,jpg,grain
portrait photo of a german man outside a night club, morning, 1990, tuxedo, suit, <lora:badquality_v02:1> bad quality
推奨ネガティブプロンプト
(depth of field)
black and white, (depth of field)
推奨パラメータ
samplers
steps
cfg
resolution
other models
ヒント
Use '(depth of field)' as a negative prompt to greatly improve the realistic look.
バージョンのハイライト
I actually cant remember what extra I did, but I know this is trained on a bunch of extra crappy pics. Works better than v01.
as title says, this is designed to mess up the image in the pursuit of realism. It's intended tol break your whitebalance, add grain, and blur and other organic things Ai avoids usually.
Help it along with some "bad quality, jpg, grain" type prompts.
(depth of field) in the negative helps a LOT for the look
Updated with a v02 that works a bit better. Can even function with minimal prompt help now.















