update metadata
#1
by prithivMLmods - opened
README.md
CHANGED
|
@@ -1,3 +1,67 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
base_model:
|
| 6 |
+
- Qwen/Qwen-Image-Edit-2511
|
| 7 |
+
pipeline_tag: image-to-image
|
| 8 |
+
library_name: diffusers
|
| 9 |
+
tags:
|
| 10 |
+
- lora
|
| 11 |
+
- art
|
| 12 |
+
- Studio-DeLight
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+

|
| 16 |
+
|
| 17 |
+

|
| 18 |
+
|
| 19 |
+
# **QIE-2511-Studio-DeLight**
|
| 20 |
+
|
| 21 |
+
> **QIE-2511-Studio-DeLight** is an adapter LoRA developed for Qwen’s **Qwen-Image-Edit-2511** image-to-image model, designed to relight images with neutral, studio-style illumination. The model removes existing lighting and replaces it with soft, evenly diffused light while preserving the original identity, composition, textures, and geometry. It ensures balanced exposure, realistic shading, and consistent results without artifacts, harsh highlights, or directional shadows.
|
| 22 |
+
|
| 23 |
+
## **Quick Start with Diffusers**
|
| 24 |
+
|
| 25 |
+
> Compatible with versions **2509** and **2511**.
|
| 26 |
+
|
| 27 |
+
```bash
|
| 28 |
+
pip install -U diffusers transformers accelerate
|
| 29 |
+
```
|
| 30 |
+
|
| 31 |
+
```python
|
| 32 |
+
import torch
|
| 33 |
+
from diffusers import DiffusionPipeline
|
| 34 |
+
from diffusers.utils import load_image
|
| 35 |
+
|
| 36 |
+
# Switch to "mps" for Apple devices
|
| 37 |
+
pipe = DiffusionPipeline.from_pretrained(
|
| 38 |
+
"Qwen/Qwen-Image-Edit-2511",
|
| 39 |
+
dtype=torch.bfloat16,
|
| 40 |
+
device_map="cuda"
|
| 41 |
+
)
|
| 42 |
+
|
| 43 |
+
pipe.load_lora_weights("prithivMLmods/QIE-2511-Studio-DeLight")
|
| 44 |
+
|
| 45 |
+
prompt = "Neutral uniform lighting Preserve identity and composition"
|
| 46 |
+
|
| 47 |
+
input_image = load_image(
|
| 48 |
+
"https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png"
|
| 49 |
+
)
|
| 50 |
+
|
| 51 |
+
image = pipe(image=input_image, prompt=prompt).images[0]
|
| 52 |
+
```
|
| 53 |
+
|
| 54 |
+
**Trained on [ModelScope.cn](https://modelscope.cn)**
|
| 55 |
+
|
| 56 |
+
## **Trigger Prompt**
|
| 57 |
+
|
| 58 |
+
Use the following prompt to activate the studio relighting behavior:
|
| 59 |
+
|
| 60 |
+
```
|
| 61 |
+
Neutral uniform lighting Preserve identity and composition
|
| 62 |
+
```
|
| 63 |
+
|
| 64 |
+
## **Download Model**
|
| 65 |
+
|
| 66 |
+
You can download the model files from the **Files & versions** tab:
|
| 67 |
+
[Download](https://huggingface.co/prithivMLmods/QIE-2511-Studio-DeLight/tree/main)
|