Activation tag: maomi jita
v1 z-image should be used with strength 2 on turbo, and all examples were generated on turbo despite civit saying it's base.
---
I have been enlightened.
v0.5 z-image, trained on Z-turbo, is overcooked, prone to body horrors and incoherent backgrounds.
v1 z-image, trained on Z-base, is better at all aforementioned. However it lost sovl and I find it boring artistically.
I'd recommend v0.5 tbh

