README.md

---
license: apache-2.0
base_model:
- Qwen/Qwen-Image-Edit-2509
pipeline_tag: image-to-video
tags:
- lora
language:
- en
library_name: diffusers
---
# Model Card for Qwen Edit Image 2509 LoRA "This Person"

Public version of "This Person" Lora for Qwen Edit Image 2509.

I found LoRA strength from 1 to 1.3 to work as it was intended.

I use Euler/Normal, 12 steps with Lightning Lora 4-steps.

Use this color to mask the placement of the person #56FF00

![Example](ref.png)

Begin your prompt with "This person" and describe what they wear, what they're doing.

You can specify what hairstyle they have and other attributes. If you don't specify their clothing, they will be depicted exactly as on the reference image.

You don't need to prompt the lighting conditions. It automatically adapts.


Make sure the face is visible as close as possible.
If you'll use a full-body shot of the person - and you don't get the consistency - try to add a close-up of their face into the reference like this:

![Example](ref_1.png)

Here are sone prompt examples:

"This person lies in bed covered up to her chin with a gray plaid blanket. Her head rests on a pillow as she looks toward the side with a weary expression on her face."

"This person wears a striped work shirt. He stands, holding a black telephone to his ear with one hand. His other hand rests out of side as he listens intently, his expression serious and thoughtful."

"This person has short dark hair and is visible from the back. He wears a brown coat, dark trousers and black shoes. He stands, facing away from the camera. His posture is still and contemplative."

Turned out you can use this LoRA also to put objects on their places. So far I tried "This car", "This toy"...Please experiment and enjoy!