Loading
Vous avez un workflow ComfyUI ? Obtenez une documentation lisible directement dans votre navigateur.Rendre un workflow lisible.Essayer maintenant ↗
Another somewhat successful embed. Like Krile, this one works mostly on Anime checkpoint, as it was auto-tagged by WD1.4 and trained on NAI - booru tag training will thus give the best results.
Source du modèle
Description de la source
Another somewhat successful embed. Like Krile, this one works mostly on Anime checkpoint, as it was auto-tagged by WD1.4 and trained on NAI - booru tag training will thus give the best results.
It does give some pretty close looks in photorealism though! I may, however, use this lora to make a more realistic data set and train a second version for that.
Second caveat - this dataset of fanart and screenshots didn't really capture the back of her, so her skirt/bustle won't show up, and back shots will be mostly nude as a result as well - even if the dataset was fully SFW.
Other than that, hope you enjoy!
Sources
1 sourceArtefacts du modèle
1 artefactExtraits de sources
1 extraitSource context: 240 downloads · Type TextualInversion · Base model SD 1.5
