To avoid recurring questions like: "The photo looked better before removing the lights, what's the point of this trash LoRa?"

First, I would like to clarify that if you have ever trained a LoRa lighting dataset, you know how difficult it is to remove lighting from human faces. However, training requires pairs of images (one with lighting, one without).

For example, the previous method for creating a dataset was to first find a photo without lighting, and then use a LoRa lighting model to add lighting.

This approach has two drawbacks: 1. Photos without lighting are hard to find. 2. AI-added lighting is never as natural as real photos and requires constant trial and error.

Now, using this LoRa, you can directly find any model photo featuring your desired lighting effect and use this LoRa to remove the lighting. The resulting LoRa lighting dataset is significantly improved compared to previous ones.

Here is a user guide:

YouTube: https://youtu.be/ReKNKBQQky4

Bilibili: https://www.bilibili.com/video/BV1Tc11BJETN/

Fixed prompt (no modification needed):

" 移除光影,使用柔和光线对图片进行重新照明"

Instructions: Download the LoRA file to the models/loras folder.

You also need this LoRA module and should use them together: https://huggingface.co/lightx2v/Qwen-Image-Lightning/tree/main

For communication/collaboration, you can join the Discord group: https://discord.gg/yVAVa43mWk

If you find these resources helpful, or if you use them for commercial purposes, buy me a coffee! Thank you for supporting original content! PayPal account: Daniel8152