license: apache-2.0

Overview

These LoRAs are experiments designed to speed up finetuned Z-Image base model.

How they were made

These LoRAs were created by extracting the weights difference between Z-Image Base and Z-Image-Distilled.

Credits

Special thanks to GuangyuanSD for providing the distilled model.

Recommended inference parameters

inference cfg: 1.0โ€“2.5 (recommended range: 1.0~1.8; higher values enhance prompt adherence)
inference steps: 10โ€“20 (10 steps for quick previews, 15โ€“20 steps for more stable quality)
sampler / scheduler: Euler / simple, or res_m, or any other compatible sampler

Comparisons:

Finetuned model used: Z-Image Spice (Experimental)

lora-disitill-test-lora_2

Ranks:

LoRAs at strength 1.0

lora-disitill-test-rank_

Model: Z-Image Spice (Experimental)

Prompt positive:

A vibrant and detailed photograph capturing the iconic atmosphere of "Moulin Rouge." The central figure is a woman dressed in an elaborate red and blue corseted gown with intricate lace detailing. She stands confidently with her hands on her hips, exuding elegance and allure. Her hair is styled in an ornate updo adorned with feathers and flowers, complementing the vibrant colors of her attire.

In the background, a bustling scene unfolds under a night sky. The cobblestone street is lined with people dressed in period costumes, suggesting a lively, historical setting. To the left, a large sign reads "Moulin Rouge" in bright lights, indicating the location's name and charm. On the right side of the image, another prominent sign displays part of the word "RA..." which likely represents "Rajah," hinting at additional entertainment venues.

The woman is wearing jewelry including earrings and a necklace that adds to her glamorous appearance. The background also features elements such as buildings with illuminated signs, street lamps casting a warm glow, and people in formal attire, reinforcing the upscale and festive atmosphere of Moulin Rouge.

Prompt negative:

3d render, cartoon, lowres, low resolution, blurry, watermark, signature, messy lighting, double chin, over-sharpened, high contrast

Simple | Euler | 422758183625811

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support