Comparison of image and face fusion effects on various platforms

Face fusion is defined by many people as face-changing. In fact, strictly speaking, face fusion and face-changing are different. Face fusion is the fusion of two faces, based on the key points of face detection to form a new face, but the new face picture is similar to the original two face pictures.

Face-changing is the swapping of human faces. The resulting output face picture will only resemble one of the input faces. Therefore, based on this characteristic, face fusion and face-changing are two different things. meaning.

Below we will compare the face fusion effects of multiple platforms currently on the market, as well as their pros and cons.

Ali

advantage:

1. No restrictions on imported templates
2. Composite effect is relatively better

 

 

shortcoming:

1. The effect of beard fusion needs to be optimized

 

 

baidu

advantage:

1. You can specify the face in the picture

shortcoming:

1. There are restrictions on the templates that can be imported.
2. The face in the template image is not very good due to partial fusion
3. It also handles beards poorly.

 

 

Tencent

advantage:

1. Support the fusion of multiple faces, for example, no more than 6 faces in one picture
2. Face swap + face fusion
3. The similarity of face fusion can adjust the parameters

 

shortcoming:

1. The face in the template image is not very good due to partial fusion
2. It also handles beards poorly.
3. There is a limit to the number of templates

Face++

advantage:

1. The effect of fusion is better

 

 

shortcoming:

1. There are also some flaws in the processing of the beard
2. There will be missing or excess for the fusion of hair