When using it, be sure to select photos with relatively consistent faces for replacement. The effect will be better, such as swapping front-facing photos with front-facing ones, or side-facing photos with side-facing ones.

The default is to use head-to-head replacement. If the face photo and reference photo do not contain a complete head, you can only choose to input face.

100 retains the face, redraws the area outside the face to the head frame, and retains all background areas outside the head. If the head frame expansion area results in poor face fusion in the output image, you can increase this value. If an error occurs, it means the value exceeds the image's limits. Especially if the head is at the edge of the image, try not to modify this value too much. If the head is in the middle of the image, this value can be increased.

If you're unsure about the head number to replace in a group photo, you can use this workflow to query the head number first and then gradually replace the face.https://www.runninghub.cn/ai detail/1942616984574210049