Currently, the Z-Image base model is producing some poor-quality results with this model. I’ve been trying to improve it, but I’d really appreciate your feedback on which aspects of the model need further work and improvement.
(NSFW support will be available in a future update, so stay tuned! This model is still in development and actively being refined.)
Recommended Parameters
Based on my testing, the best results were achieved with:
Steps: 30
CFG: 2.5
Sampler: Res Multistep
Scheduler: Simple
You can experiment with these settings to find what works best for you!


.jpeg)