Hi EvelynFan,
I noticed that VOCA's own dataset has a frame rate of 60 frames, but in Faceformer, both training and inference have set the frame rate to 30 frames. Why is this done? Also, if the frame rate is different, how is it quantitatively compared with the VOCA method? Is it aligned by downsampling the frame rate? Thank you for your help!
Hi EvelynFan, I noticed that VOCA's own dataset has a frame rate of 60 frames, but in Faceformer, both training and inference have set the frame rate to 30 frames. Why is this done? Also, if the frame rate is different, how is it quantitatively compared with the VOCA method? Is it aligned by downsampling the frame rate? Thank you for your help!