AI One-click Group Photo ComfyUI

The image is from the internet, delete if infringing

The purpose of this workflow

One-click group photo workflow

Operating configuration RTX 3060

Runtime is approximately 400 seconds, if veering off topic, restart.

This workflow contains multiple node groups

1. Use YoloWorld to partition the male and female areas and background regions (choose a group photo with a clear distinction between male and female, if it is a same-sex workflow, there are annotations)

2. Automatically generate red, green, and blue mask layers.

3. Use IPA to partition and draw CN to control character status.

4. Analyze uploaded group images and generate prompt words using a large language model, the large language model uses llava.

Model download:https://ollama.com/library  Open CMD and run ollama run llava.

5. Finally use Reactor to swap faces and improve simulation accuracy.

6. Use depth map to repair hands (if the repair feels odd in actual use, you can reduce the noise)

7. Finally, you can individually useSUPIR to enlarge and add details, I personally like to use this, although it's slower, but the V2 version is much faster.

Upload three images, one group photo, one male photo, one female photo, click run to start.