Introduction

Latent Bridge Matching (LBM) is a new, versatile and scalable method proposed in LBM: Latent Bridge Matching for Fast Image-to-Image Translation that relies on Bridge Matching in a latent space to achieve fast image-to-image translation. This model was trained to relight a foreground object according to a provided background.

https://huggingface.co/jasperai/LBM_relighting

https://github.com/kijai/ComfyUI-LBMWrapper

https://huggingface.co/jasperai/LBM_relighting/blob/main/model.safetensors

Recommended machine：Large-PRO

Workflow Overview

How to use this workflow

Step 1: Load Image

Upload your portrait or pet at the top and the background image at the bottom

Step 2: Adjust Image parameters

The scale on the left is used to adjust the overall resolution of the image, and the LBMSampler on the right is mainly used to restore details. Tests show that there is basically no change after more than 30 steps

Step 3: Adjust the mask

kernel_size is used to adjust the blur of the mask, and expand is used to adjust the shrinkage of the mask

Step 4: Re-cover the portrait

In order to further eliminate the extra early light generated by LBM_Relighting, the mask of the original image is used here to cover the portrait features again in the output

Step 5: Detail Repair

Re-cover the secondary acquired portrait mask in the output for feathering

Step 6: Partial Redrawing

Use Flux Fill to adjust the output later, such as adding portrait shadows