# Layer Extract & Layer Remove Loras For Qwen Image 2.1 — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/eb14c1c48b3c8cb9

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-09-28 12:08 ingested 2026-09-29 02:40

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1wsctqz/layer_extract_layer_remove_loras_for_qwen_image_21/)

## retained text



```
✂️ LayerExtract isolates a prompt-specified subject and exports it on a transparent background for independent editing. 

[https://modelscope.ai/models/DiffSynth-Studio/Qwen-Image-2.1-LayerExtract](https://modelscope.ai/models/DiffSynth-Studio/Qwen-Image-2.1-LayerExtract)

✨ LayerRemove takes the source image and a reference layer, removes the corresponding object, and reconstructs the scene behind it. 

[https://modelscope.ai/models/DiffSynth-Studio/Qwen-Image-2.1-LayerRemove](https://modelscope.ai/models/DiffSynth-Studio/Qwen-Image-2.1-LayerRemove)

⚡ Both are compact Qwen-Image-2.1 LoRAs that can be hot-swapped within the same DiffSynth-Studio pipeline. 

🧩 Together they turn flattened images into editable assets for compositing, repositioning, replacement, and design workflows. 

📜 LoRA weights: Apache 2.0. Qwen-Image-2.1 base-model terms also apply.

  

```



document fetched 2026-09-29 00:32 · canonical https://reddit.com/r/StableDiffusion/comments/1wsctqz/layer_extract_layer_remove_loras_for_qwen_image_21
