NEW ERA v3.0 training and others
I have significantly expanded, improved and finalized my retro model, it is now in training. I've been trying to get the perfect fine tuning of my "new era v2.1" for the last few days, it took an awful lot of time, tons of reading and thousands of attempts, I'm still not 100% happy with the result, but I'm getting close. From what happened: the eyes, hands, image volume, background detail have improved greatly, but one drawback outweighs all the advantages - the model turns out to be simply terribly “overcooked” (contrasting and bright); Until I solve this problem at least partially, I won’t be able to post it. But expect a major upgrade in quality soon.
All images are taken at extremely low configuration scale settings, which affects anatomy and response to cues.
My AIRL lora is almost ready too (anime in real life), stay tuned)
news
