---
tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
widget:
- output:
    url: images/1.png
  text: 'Synthetic Face, a close-up shot of a young mans face features a maroon baseball cap adorned with a leather band. The mans hair is cut short and neatly trimmed. His eyes are a piercing blue, and his eyebrows are a darker shade of brown. He is wearing a gray tank top with a silver chain around his neck, adding a pop of color to his chest. The backdrop is a textured gray wall.'
- output:
    url: images/2.png
  text: 'Synthetic Face, a beautiful blonde woman with long, wavy blonde hair stands in front of a dark gray backdrop. She is dressed in a red strapless dress, adorned with silver earrings. Her lips are painted a vibrant red, adding a pop of color to her face. Her eyes are a piercing blue, and her eyebrows are a darker shade of brown. Her hair is cascading down her shoulders, framing her entire face.'
- output:
    url: images/3.png
  text: 'Synthetic Face, a medium-sized man stands in front of a stark white backdrop. He is dressed in a black tuxedo, adorned with a white collared shirt and a black bow tie. His eyes are a deep blue, and his hair is a rich black, adding a pop of color to the scene. His lips are a lighter shade of pink, and he has a slight smile on his face. His eyebrows are a darker shade of blue, adding depth to the composition.'
base_model: Qwen/Qwen-Image
instance_prompt: Synthetic Face
license: apache-2.0
---

![1.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/BkJ7u56OLwhhKYI1VIPxg.png)

# Qwen-Image-Synthetic-Face

<Gallery />

---

# Model description for Qwen-Image-Synthetic-Face

Image Processing Parameters 

| Parameter                 | Value  | Parameter                 | Value  |
|---------------------------|--------|---------------------------|--------|
| LR Scheduler              | constant | Noise Offset              | 0.03   |
| Optimizer                 | AdamW  | Multires Noise Discount   | 0.1    |
| Network Dim               | 64     | Multires Noise Iterations | 10     |
| Network Alpha             | 32     | Repeat & Steps           | 22 & 2650 |
| Epoch                     | 20  | Save Every N Epochs       | 2  |

    Labeling: florence2-en(natural language & English)
    
    Total Images Used for Training : 26 [HQ Images]

## Data Sources

| Source       | Link                                |
|--------------|-------------------------------------|
| Playground   | [playground.com](https://playground.com/) |
| ArtStation   | [artstation.com](https://www.artstation.com/) |
| 4K Wallpapers| [4kwallpapers.com](https://4kwallpapers.com/) |

## Best Dimensions & Inference

| **Dimensions** | **Aspect Ratio** | **Recommendation**       |
|-----------------|------------------|---------------------------|
| 1472 x 1140      | 4:3 (approx.)   | Best                     |
| 1024 x 1024     | 1:1              | Default                  |

### Inference Range

- **Recommended Inference Steps:** 35-50

## Setting Up
```python
import torch
from diffusers import DiffusionPipeline

base_model = "Qwen/Qwen-Image"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "prithivMLmods/Qwen-Image-Synthetic-Face"
trigger_word = "Synthetic Face"  
pipe.load_lora_weights(lora_repo)

device = torch.device("cuda")
pipe.to(device)
```

## Trigger words

You should use `Synthetic Face` to trigger the image generation.


## Download model


[Download](/prithivMLmods/Qwen-Image-Synthetic-Face/tree/main) them in the Files & versions tab.