General scenario: Detect and align the face angles of the first frame of two images (source/target).
Because the face and key points of the face have been captured,
Guess: In fact, the program should be able to determine the initial angle difference of the two heads?
(1) By comparing the initial head difference values (turning angle, eye/mouth opening, etc.) of (input source image/target video first frame),
(2) Then use the program to adjust the input source image to keep it consistent with the initial angle of the first frame of the target video.
This helps to quickly get the best result;
there is no need to manually align the first frame or select the source image portrait angle every time.
I hope someone is interested in this, and it would be best if it can be implemented?
大致情景：检测并对齐（源/目标）两个画面的第一帧人脸角度。
因为已经捕获到面部及面部的关键点信息， 猜测：实际上程序应该能判断两个头部的初始角度差异？
（1）通过对比（输入源图片/目标视频第一帧）的初始头部的差异值（转向角度，眼睛/嘴巴张合等）， （2）然后用程序的方式，调节输入源图像，与目标视频的第一帧初始角度，保持一致性。
这样有利于快速得到最优结果； 而不必每次在第一帧都要进行人工对齐，或挑选源图片人像角度。
希望有人对此感兴趣，最好能够实现它？
General scenario: Detect and align the face angles of the first frame of two images (source/target).
Because the face and key points of the face have been captured, Guess: In fact, the program should be able to determine the initial angle difference of the two heads?
(1) By comparing the initial head difference values (turning angle, eye/mouth opening, etc.) of (input source image/target video first frame),
(2) Then use the program to adjust the input source image to keep it consistent with the initial angle of the first frame of the target video.
This helps to quickly get the best result; there is no need to manually align the first frame or select the source image portrait angle every time.
I hope someone is interested in this, and it would be best if it can be implemented?