looking at the previews, this lora might be the best one to one style

Firstly, I wanna say - fantastic job training this; it really captures the style perfectly.

I'm having some trouble replicating the example images you've provided in the current version of Forge, though, even with copying the generation parameters straight out of the images via PNG Info. I've messed around with a bunch of the variable settings that don't always show in metadata (scheduler, ENSD, SGM noise mult, discarding penultimate sigma, downcasting model alphas, Forge's external software replication options, etc.) but I haven't had any luck. The images are always fairly similar, but not exactly "it", so I'm wondering if you have any particular configuration options in your environment that I'm missing?

Its greatest