Talking-head-Generation-with-Rhythmic-Head-Motion
Talking-head-Generation-with-Rhythmic-Head-Motion copied to clipboard
lelechen63
Hello，what's the meaning of consider_key in the function 'RT_compute'?
By the way, could you explain how to normalize 3d landmark more specific? Thanks very much!!
'''python
def RT_compute(video_path = None, path = None): #video path should be the original video path
consider_key = [1,2,3,4,5,11,12,13,14,15,27,28,29,30,31,32,33,34,35,39,42,36,45,17,21,22,26]
source = np.zeros((len(consider_key),3))
ff = np.load('../basics/standard.npy')
for m in range(len(consider_key)):
source[m] = ff[consider_key[m]]
source = np.mat(source)
if video_path != None:
lmark_path = video_path[:-4] + '__original.npy'
rt_path = video_path[:-4] +'__rt.npy'
front_path = video_path[:-4]+'__front.npy'
# normed_path = os.path.join( person_path,vid[:-12] +'normed.npy')
if os.path.exists(front_path):
print ('No front path ')
'''
