Is there a way to keep the unprocessed data?

[G-force78]

The data that is generated before the video is made? Such as the individual frames and face landmark frames?

[ZJYang]

As for video original frames, you can use ffmpeg to convert the video to images. As for pose landmark images, you can save pose_image_np here.

AniPortrait/scripts/vid2vid.py

Line 203 in 90f1b67

 pose_image_np = cv2.resize(lmk_img,  (width, height))