Open qiqiApink opened 1 year ago
Hi, at this step we aim to unify the skeleton structure of all motions. In other words, we wanna all motions to have the same scale, it is good for model learning. Here, we just randomly selected 000021 as the retargeting target.
In raw_pose_processing.ipynb, you choose example_id = "000021" and the first frame of it to calculate the target offsets. But it seems that different id and different frame will result in different target offsets. Can you tell me why you choose the first frame of "000021" but not calculate the target offsets for each frame of each id?