Model Source: https://www.modelscope.cn/models/wikeeyang/Z-Image-Turbo-Art/summary

Model Description: This model is a merged and fine-tuned version of Z-Image and Z-Image-Turbo. It extracts part of the stylization capabilities from the Z-Image base model, merges them layer-by-layer with Z-Image-Turbo, and undergoes fast fine-tuning. This is currently an attempt to fully utilize the Z-Image base model. Compared to the official original Turbo model, this model produces clearer images and stronger stylization capabilities; however, there is a slight drop in portrait quality, particularly skin delicacy, while text rendering capabilities are mostly retained. Please refer to the sample images below for specific visual results. The model is easy to use, with good parameter adaptability and LoRA compatibility. Personally, I believe this model is better suited for generating and creating artistic styles and decorative advertising images.

How to Use: Basic workflow please refer to: Z-Image-Turbo official workflow Inference CFG: 1.0; Inference steps: 8-12; Sampler / scheduler: ddim / simple, or any other. Also on: Huggingface.co, Civitai

Model License: Please follow the Apache-2.0 open source license for the Z-Image model.