Z-Image Modern Anime Model Card

This is an experimental model. I full-finetune Z-Image by Quality Tuning only.

Usage

• ComfyUI
  1. Download the model.
  2. Enjoy! (trigger word: Japanese modern anime style, )
• diffusers
  1. Run the code:

import torch
from diffusers import ZImagePipeline

# Load the pipeline
pipe = ZImagePipeline.from_pretrained(
    "alfredplpl/z-image-modern-anime",
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=False,
)
pipe.to("cuda")

# Generate image
prompt = "Japanese modern anime style, an upper body shot of an woman standing on the rainy street. She watches the sky. "
negative_prompt = "photo, cg, 3d, blurry"

image = pipe(
    prompt=prompt,
    negative_prompt=negative_prompt,
    height=1280,
    width=720,
    cfg_normalization=False,
    num_inference_steps=30,
    guidance_scale=4,
    generator=torch.Generator("cuda").manual_seed(42),
).images[0]

image.save("example.png")