Introduction
The face replacement workflow built based on ACE++'s lora model and flux-dev brings users an unprecedented face replacement experience with its precise feature analysis, efficient data interaction, and excellent image optimization capabilities. Whether it is used for creative photography, film and television special effects production, or interesting social sharing, this workflow can be your right-hand man.
ACE++:
ACE++ is a post-training model based on the FLUX.1-dev series of Black Forest labs. It is a lora. It can be used in conjunction with the Flux Fill model without installing a plug-in. We can use the comfyui_local_lora16.safetensors model to perform face replacement tasks. Just download the model and put it in the lora folder. The usage is the same as lora.
https://huggingface.co/ali-vilab/ACE_Plus/tree/main/portrait
Workflow Overview
How to use this workflow
Step 1: Insert image resources
Load the face image in the red box above,Load the image to be face-swapped in the box below.
Step 2: Loading ACE++ lora model
Put the downloaded ACE++ lora model into the red box and adjust it to the weight that suits you.
Step 3: Facial expression regulation
Here you can adjust the task's smiling level, winking and a range of other actions.
Step 4: FaceDetailer
In the red box, you can adjust the degree of facial refinement until you are satisfied.
Summarize
As a new generation of artificial intelligence open platform, ACE++ provides comprehensive face recognition technology. These functions are widely used in the business field, especially in scenarios that require highly personalized services. Let's create together!