Z-Image Multi-functional Base Model User Guide

1. Base Model Features

An AI generation model tailored for young female themes that balances realistic texture and aesthetic atmosphere. Its core advantages are as follows:

Image Quality Performance

Built on the Z-Image-Turbo technical framework, it delivers soft-style portrait outputs while preserving intricate details of features and hair strands, alongside realistic light and shadow layers. The artwork achieves a perfect blend of atmospheric vibe and authentic realism.

Style Characteristics

Features warm soft light and low-saturation color matching. It excels at creating diverse gentle temperaments such as cute style and mature soft elegance. It stably generates high-quality works for daily portrait photography and relaxed home-style looks, presenting natural and lifelike character textures.

Extended Capabilities

Compatible with the Z-Image-Edit function. It supports creative image conversion via bilingual editing commands, enabling flexible adjustment of styling, scenes and other elements while balancing creative freedom and stylistic consistency.

LoRA Introduction: Sheer Blue Cheongsam Girl Portrait

A dedicated LoRA trained for creating sheer new Chinese-style portraits of young girls in blue cheongsams. It perfectly restores the gentle Chinese-style ambiance of sunny blue sky + pale blue cheongsam + butterfly hair accessories, adapting to clean and translucent soft light effects to effortlessly produce ethereal portrait shots with a natural sense of vitality.

Core Adaptable Styles: Sheer pure allure style, new Chinese-style portrait photography, outdoor blue sky portrait

Costume & Accessories: Pale blue jacquard cheongsam / modified cheongsam, blue-and-white butterfly hair ornaments, pearl and jade jewelry, sheer manicure

Atmosphere Matching: Sunny blue sky backdrop, soft natural light, rim lighting on hair, clean and translucent skin texture

✨ Recommended Trigger Words (Copy & Paste Directly)

Positive Prompt:

masterpiece, best quality, 1girl, solo, young girl, clean and fresh makeup, light blue cheongsam, delicate butterfly hairpin, blue sky background, soft natural light, clear skin texture, wind blowing hair, soft focus, cinematic light, new Chinese style, photo-realistic portrait, pure and ethereal vibe

2. Usage Instructions

(1) Online Operation

Main Subject Generation: Although it supports generating works with the "1girl" main subject, its performance in processing long text descriptions is far superior. Its stability for simple "1girl" generation is slightly insufficient, so this method is not highly recommended.

Negative Tags: It is recommended to use the default negative tags on Liblib Platform directly, including ng_deepnegative_v1_75t, (badhandv4:1.2), EasyNegative, (worst quality:2). These tags optimize generated image quality and avoid unnecessary visual flaws.

(2) Local Deployment

Workflow & Weight: Follow the workflow shown in the reference images and set the weight to 0.9 — an ideal parameter verified through repeated testing for optimal generation results.

Hardware Requirements: To ensure generation precision and quality, set the sampling steps to 30 steps or more. For local deployment, a graphics card with 12GB+ VRAM is recommended to provide sufficient computing power and avoid degraded output due to hardware limitations.

3. Usage Restrictions

(1) Platform Hosting Restrictions

It is strictly prohibited to host this base model and its derivative versions (e.g., merged models) on any websites or applications for profit or donation purposes.

For commercial use (including generative service provision, image sales, short video article creation, e-commerce design, publication production, etc.), please contact the author via official channels to obtain official authorization.

(2) Image Sales Rules

Direct sales of images generated by this base model are strictly forbidden.

If you intend to sell such images, you must conduct in-depth manual revisions to make the works legally recognized as completely original creations. Any legal liabilities arising from violations shall be borne solely by the user; the author assumes no relevant responsibilities.

(3) Content Creation Specifications

Users shall not utilize this base model to intentionally create, disseminate or output illegal and . Please adhere to public order and good customs, apply the model to positive and legitimate creation scenarios, and jointly maintain a healthy creative ecosystem.

(4) Model Merging Regulations

Merging operations on this base model are strictly prohibited to preserve its original characteristics and operational stability.

4. Installation Steps

(1) Forge Deployment

Base Model Placement: Put the base model file into the models\Stable-diffusion folder for system automatic recognition and invocation.

VAE Placement: Place the VAE file into the models\VAE folder (manual download required). The VAE file is critical for optimizing image color performance and detail rendering.