In Reality Scan, when using the video sequence feature, why are only the first few keyframes extracted from the video?