---
tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
widget:
- output:
    url: images/1.png
  text: >-
    Qwen Anime, An animated animated image of a man holding a sword in his right
    hand. The man is wearing a white collared shirt with a white collar. His
    left hand is raised in the air with a ring on it. His right hand is bent at
    the top of the handle of the sword. His eyes are blue and he has short black
    hair. His hair is cut in a ponytail that is falling to the right side of his
    head. The background is a vibrant blue color. To the right of the man is a
    window with a yellow door.
- output:
    url: images/2.png
  text: >-
    Qwen Anime, An animated image of a mage girl in a long purple cloak with silver runes. She has long white hair, and her blue eyes glow with magical light. Her hands are raised, summoning glowing circles of magic. The background is a floating library with levitating books.

- output:
    url: images/3.png
  text: >-
    Qwen Anime, An animated image of a young swordsman in a white robe with red trim. He has black hair tied in a top knot and carries a wooden katana. His eyes glow faintly green. The background is a peaceful bamboo forest.
base_model: Qwen/Qwen-Image
instance_prompt: Qwen Anime
license: apache-2.0
---

![1.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/xy97ap4worKPeKv1H6wv5.png)

# Qwen-Image-Anime-LoRA

<Gallery />

---

# Model description for Qwen-Image-Anime-LoRA

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           | 25 & 3000 |
| Epoch                     | 20  | Save Every N Epochs       | 1  |

    Labeling: florence2-en(natural language & English)
    
    Total Images Used for Training : 44 [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**       |
|-----------------|------------------|---------------------------|
| 1664 x 928      | 16:9 (approx.)              | Best                     |
| 1024 x 1024     | 1:1              | Default                  |

### Inference Range

- **Recommended Inference Steps:** 40-50

## 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-Anime-LoRA"
trigger_word = "Qwen Anime"  
pipe.load_lora_weights(lora_repo)

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

## Trigger words

You should use `Qwen Anime` to trigger the image generation.


## Download model


[Download](/prithivMLmods/Qwen-Anime-LoRA/tree/main) them in the Files & versions tab.