It's a merge from rather old realistic Checkpoints and it was nothing more than an experiment, I almost forgot. Just like the models, mostly my own unreleased Checkpoints and LoRas, I've been tinkering around with. One day, when there was the possibility to merge LoRas to Checkpoints, I mixed a lot of ??? and also the excursion_xLCM , created by mndtrblnc into that mess. In the end it's not Turbo-fast, but it works well with

10 inference steps,

DPM++SDE (Karras) or Euler a

1 - 2 guidance scale

and you don't need any LoRa to put ??? on the faces.

You'd rather write ???, facial, ????? into the negative prompt, if you don't want a gooey smile :o)

Of course, you can generate landscapes, cars, and further bs with this model.