---
title: Qwen-Image-2.1-LayerRemove
canonical_url: "https://www.modelscope.ai/models/DiffSynth-Studio/Qwen-Image-2.1-LayerRemove"
md_url: "https://www.modelscope.ai/models/DiffSynth-Studio/Qwen-Image-2.1-LayerRemove.md"
repository: DiffSynth-Studio/Qwen-Image-2.1-LayerRemove
last_updated: 2026-09-28
license: "Apache License 2.0"
pipeline_tag: text-to-image-synthesis
tasks:
  - text-to-image-synthesis
base_model:
  - Qwen/Qwen-Image-2.1
base_model_relation: finetune
parameters: 83.9M
tensor_type:
  - BF16
library_name:
  - pytorch
  - safetensors
supports_inference: txt2img
downloads: 9
stars: 0
---

# Qwen-Image-2.1-LayerRemove

> Qwen-Image-2.1-LayerRemove - An open-source model by DiffSynth-Studio on ModelScope. Layer Removement (Qwen-Image-2.1 LoRA)

DiffSynth-Studio/Qwen-Image-2.1-LayerRemove is a 83.9M-parameter text-to-image-synthesis model on ModelScope. licensed under Apache License 2.0. derived from Qwen/Qwen-Image-2.1. and supports online inference (txt2img).

- **Repository**: DiffSynth-Studio/Qwen-Image-2.1-LayerRemove
- **License**: Apache License 2.0
- **Tasks**: text-to-image-synthesis
- **Parameters**: 83.9M
- **Base model**: Qwen/Qwen-Image-2.1
- **Online inference**: txt2img
- **Downloads**: 9
- **Stars**: 0
- **Last updated**: 2026-09-28

Source: https://www.modelscope.ai/models/DiffSynth-Studio/Qwen-Image-2.1-LayerRemove

---

# Layer Removement (Qwen-Image-2.1 LoRA)

This model is designed to remove specific elements from an image. The removed layers are saved as images with transparent backgrounds, making them well-suited for detailed editing workflows.

* Training Framework: [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio)
* Base Model: [Qwen-Image-2.1](https://www.modelscope.cn/models/Qwen/Qwen-Image-2.1)
* Dataset: [PrismLayersPro](https://www.modelscope.cn/datasets/artplus/PrismLayersPro)
* Training Resources: 8 ZW810 PPUs
* Training Steps: 60,000 steps

Suggested prompt: `Remove the object from Figure 2 in Figure 1`

## Showcase

Prompt: `Remove the object from Figure 2 in Figure 1`

| Input Image 1 | Imput Image 2 | Output Image |
|-|-|-|
|![](./assets/image_input_2.png)|![](./assets/image_extract_girl_wings.png)|![](./assets/image_remove_girl_wings.png)|
|![](./assets/image_input_2.png)|![](./assets/image_extract_wings.png)|![](./assets/image_remove_wings.png)|
|![](./assets/image_input_1.png)|![](./assets/image_extract_girl.png)|![](./assets/image_remove_girl.png)|
|![](./assets/image_input_1.png)|![](./assets/image_extract_hat.png)|![](./assets/image_remove_hat.png)|

## Inference Code

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

```shell
git clone https://github.com/modelscope/DiffSynth-Studio.git
cd DiffSynth-Studio
pip install -e .
```

Load the models and run inference:

```python
from diffsynth.pipelines.qwen_image_21 import QwenImage21Pipeline, ModelConfig
import torch
from PIL import Image
from modelscope import snapshot_download

vram_config = {
    "offload_dtype": "disk",
    "offload_device": "disk",
    "onload_dtype": "disk",
    "onload_device": "disk",
    "preparing_dtype": torch.bfloat16,
    "preparing_device": "cuda",
    "computation_dtype": torch.bfloat16,
    "computation_device": "cuda",
}
pipe = QwenImage21Pipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="Qwen/Qwen-Image-2.1", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors", **vram_config),
        ModelConfig(model_id="Qwen/Qwen-Image-2.1", origin_file_pattern="text_encoder/model*.safetensors", **vram_config),
        ModelConfig(model_id="Qwen/Qwen-Image-2.1", origin_file_pattern="vae/diffusion_pytorch_model*.safetensors", **vram_config),
    ],
    processor_config=ModelConfig(model_id="Qwen/Qwen-Image-2.1", origin_file_pattern="processor/"),
    vram_limit=torch.cuda.mem_get_info("cuda")[1] / (1024 ** 3) - 0.5,
)
pipe.enable_lora_hot_loading(pipe.dit)

# Extract layers using prompt
snapshot_download("DiffSynth-Studio/Qwen-Image-2.1-LayerRemove", allow_file_pattern="assets/*", local_dir="data")
pipe.load_lora(pipe.dit, ModelConfig(model_id="DiffSynth-Studio/Qwen-Image-2.1-LayerRemove", origin_file_pattern="model.safetensors"))

prompt = "Remove the object from Figure 2 in Figure 1"
image = pipe(prompt, seed=0, height=1024, width=1024,
             edit_image=[Image.open("data/assets/image_input_1.png"), Image.open("data/assets/image_extract_girl.png")])
image.save("image_remove_girl.png")

pipe.clear_lora()
```