AniVerse Z is a LoRA trained to generate images using the Z-Image Turbo model while maintaining the signature style of AniVerse for SD15.
The training was carried out on 520 images at a resolution of 1024×1024, produced from the AniVerse 1.2 model and refined through a workflow featuring a custom upscale step.
The training dataset includes a diverse range of visual content, with particular emphasis on human figures of various ages and multiple environments. Captions were generated afterward using an uncensored LLM.
The AniVerse Z LoRA is also capable of producing—within controlled limits—NSFW female nude images, provided that the poses are not overly complex.
Recommended Generation Settings
Sampler: Euler
Steps: 8
CFG Scale: 1.0
Scheduler: Normal
LoRA Weight: 1.0
Optional Trigger Word: AniVerse_Z
Both the training and the distribution of the AniVerse Z LoRA are approved and exclusively licensed by Samael1976 (Samuele Bonzio) to nuaion (Davide Alidosi).
All rights reserved.