Simple and effective embedding that makes images sharper and boosts contrast a bit. Works by specifically selecting "blur" concepts and pushing images away from them (if the embedding is put into negative, ofc)....
Modellquelle
Quellenbeschreibung
The difference from V1 should be not that big - I tried not to introduce "breaking" changes. Mostly, improved overall composition a bit, but sometimes it makes fingers better as a bonus.
Quellen
1 QuelleModellartefakte
1 ArtefaktQuellenauszüge
2 AuszügeSource context: 3537 downloads · Type TextualInversion · Base model SD 1.5
Source context: 3519 downloads · Type TextualInversion · Base model SD 1.5