---
license: other
license_name: stabilityai-ai-community
license_link: LICENSE.md
language:
- en
base_model:
- stabilityai/stable-diffusion-3.5-medium
pipeline_tag: text-to-image
---

<div align="center">

**Bokeh_Line_Control**

<img src="show.jpg"/>
</div>

## Description

- Input Image: Black and white line art or images processed with edge detection algorithms
- Output Image: Generated images incorporating edge information

This enables better control over the main subject features and composition, resulting in more vivid and realistic images.

## Example
| input | output | Prompt |
|---|---|---|
| ![input](./images/001_line.png) | ![output](./images/001.png) | A cat looks up, close-up, sapphire eyes |
| ![input](./images/002_line.png) | ![output](./images/002.png) | heron bird standing, closeup, graceful |
| ![input](./images/003_line.png) | ![output](./images/003.png) | a modern build design |
| ![input](./images/004_line.png) | ![output](./images/004.png) | A old woman talking |

## Use 
We recommend using ComfyUI for local inference
![input](./comfy.png)

# With Bokeh
```python
import torch
from diffusers import StableDiffusion3ControlNetPipeline
from diffusers import SD3ControlNetModel
from diffusers.utils import load_image

controlnet = SD3ControlNetModel.from_pretrained("tensorart/Bokeh_Depth_Control")
pipe = StableDiffusion3ControlNetPipeline.from_pretrained(
    "tensorart/bokeh_3.5_medium",
    controlnet=controlnet
)
pipe.to("cuda", torch.float16)

control_image = load_image("https://huggingface.co/tensorart/Bokeh_Line_Control/resolve/main/images/001_line.png")
prompt = "A cat looks up, close-up, sapphire eyes"
negative_prompt ="anime,render,cartoon,3d"
negative_prompt_3=""

image = pipe(
    prompt, 
    num_inference_steps=30,
    negative_prompt=negative_prompt, 
    control_image=control_image, 
    height=1728,
    width=1152,
    guidance_scale=4,
    controlnet_conditioning_scale=0.8
).images[0]
image.save('image.jpg')
```

## Contact
* Website: https://tensor.art  https://tusiart.com
* Developed by: TensorArt
* Api: https://tams.tensor.art/