Curiouser版本v2.0 (ID: 598451)

A merge of Bizarre v1 and Experience v10. It's good at making high-constrast images with deep shadows. It's a bit chaotic and weird, but that was intentional.

I've tested it with characters, backgrounds, and full scenes and been happy with my results. It tends to create darker images, but lighter images are also completely possible.

It does weird things with vehicles like stacking them on top of each other, which I'll need to test more. This model is still pretty experimental and I might merge in some other things to refine it in the future. I haven't tested it with anime or illustration styles yet

Faetastic makes a wonderful refiner with the swap point set at about .86 (I personally prefer the old refiner swap method, by number of steps)

描述:

I've mixed in a few new models to change things up a little bit. The model should now be better at nudity and porn (but still not perfect—use LORA's!) if you so desire, but it is still primarily safe-for-work by default (Art & Eros: A Tribute to Beauty). It will also have better contrast for dark images (Noise Offset For True Darkness in SD minus Base SD1.5)

I wanted to improve the model and make adjustments, but keep it still true to the original. I think I've succeeded

训练词语: