<h1 id="miamijay-or-multimodal-image-lora">MiamiJay 👟🌴 | MultiModal Image LoRA</h1><h3 id="premium-sdxl-x-krea-2-versions-here:">Premium SDXL x Krea 2 Versions here:</h3><a target="_blank" rel="ugc" href="https://www.kofi.com/tensorvizion">https://www.kofi.com/tensorvizion</a><a target="_blank" rel="ugc" href="https://www.patreon.com/tensorvizion/shop">https://www.patreon.com/tensorvizion/shop</a><h2 id="recommended-settings">Recommended Settings</h2>Base Model<ul><li>Qwen Image / Flux.2 Klein 9b</li></ul><h3 id="model-qwen-image-2512">Model Qwen Image 2512</h3>LoRA Weight<ul><li>0.8 (recommended)</li><li>Works well between 0.6–1.0</li></ul>Sampling Steps<ul><li>25–35</li></ul>CFG / Guidance<ul><li>4–6</li></ul>Image Size<ul><li>512x768</li></ul><h3 id="model-flux.2">Model Flux.2</h3>LoRA Weight<ul><li>0.5 (recommended)</li><li>Works well between 0.3–0.6</li></ul>Sampling Steps<ul><li>24</li></ul>CFG / Guidance<ul><li>6</li></ul>Image Size<ul><li>512x768</li></ul><h3 id="model-z-image">Model Z-Image</h3>LoRA Weight<ul><li>0.85 (recommended)</li><li>Works well between 0.7–1.2</li></ul>Sampling Steps<ul><li>28</li></ul>CFG / Guidance<ul><li>7</li></ul>Image Size<ul><li>512x768</li></ul>More base model versions will be added in the coming days, if there's one you want that I don't have message me and I'll see what I can do!Prompt Tips Describe clothing in detail for the best results. Include materials, colours, footwear, accessories, lighting, and camera style.<h3 id="example-prompt">Example Prompt</h3>&lt;jordans&gt; stylish young man wearing oversized cream hoodie, black cargo pants, premium white sneakers, luxury streetwear, layered silver chains, modern fashion photography, city streets at golden hour, ultra detailed clothing, realistic fabrics, editorial quality<h3 id="negative-prompt">Negative Prompt</h3>blurry, low quality, deformed clothing, bad anatomy, duplicated limbs, watermark, text, logo, oversaturated, poorly fitted clothes, cropped, artifacts

MiamiJay

Trained on Qwen Image 2512Thank you to ModelScope &amp; DiffSynth Studio for the training script.

Shoes_c1-st3000.safetensors

Qwen 2512

Qwen Image 1

Z-Image Vaporwave

Flux.2 Klein 9B


# 

## Model Introduction

This model is trained using the [Civision/model training](https://modelscope.ai/civision/modelTraining) environment and computing resources provided by ModelScope.

* Model Type: LoRA
* Base Model: [Qwen/Qwen-Image-2512](https://modelscope.ai/models/Qwen/Qwen-Image-2512)
* Training Code: [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio)
* Data amount：18
* Total training steps：3000
* License: Apache-2.0

## Inference Code

Install [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio):

```bash
pip install diffsynth
```

Inference:

```python
from diffsynth.pipelines.qwen_image import QwenImagePipeline, ModelConfig
import torch

pipe = QwenImagePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="Qwen/Qwen-Image-2512", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors"),
        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors"),
        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
    ],
    tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
)
pipe.load_lora(pipe.dit, ModelConfig(model_id="tensorvizion/Shoes", origin_file_pattern="Shoes_c1-st3000.safetensors"))
prompt = "a cat"
image = pipe(prompt)
image.save("image.jpg")
```


Files with SHA256: dc45536079d6f03af7b7166788d09402c2bafa44cc764d30a42f7e84e44771c9