github/DeepFaceLab
1
0
Fork 0
mirror of https://github.com/iperov/DeepFaceLab.git synced 2025-08-19 13:09:56 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'shadow_aug' of https://github.com/MachineEditor/DeepFaceLab

Browse source
This commit is contained in:
seranus 2021-12-07 21:25:13 +01:00
commit b0438ce5e9
1 changed files with 47 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

47
samplelib/SampleProcessor.py
View file

@ -266,6 +266,53 @@ class SampleProcessor(object):
img_v = np.clip (img_v + (rnd_state.random()-0.5)*a, 0, 1 ) img_v = np.clip (img_v + (rnd_state.random()-0.5)*a, 0, 1 )
img = np.clip( cv2.cvtColor(cv2.merge([img_h, img_s, img_v]), cv2.COLOR_HSV2BGR) , 0, 1 ) img = np.clip( cv2.cvtColor(cv2.merge([img_h, img_s, img_v]), cv2.COLOR_HSV2BGR) , 0, 1 )
random_shadow_amount = 1
if random_shadow_amount != 0 and rnd_state.random() < 0.5:
high_ratio=(1,2)
low_ratio=(0.01, 0.5)
left_low_ratio=(0.4,0.6)
left_high_ratio=(0,0.2)
right_low_ratio=(0.4,0.6)
right_high_ratio = (0,0.2)
#check
img = np.clip(img*255, 0, 255).astype(np.uint8)
w, h, _= img.shape
high_bright_factor = rnd_state.uniform(high_ratio[0], high_ratio[1])
low_bright_factor = rnd_state.uniform(low_ratio[0], low_ratio[1])
left_low_factor = rnd_state.uniform(left_low_ratio[0]*h, left_low_ratio[1]*h)
left_high_factor = rnd_state.uniform(left_high_ratio[0]*h, left_high_ratio[1]*h)
right_low_factor = rnd_state.uniform(right_low_ratio[0]*h, right_low_ratio[1]*h)
right_high_factor = rnd_state.uniform(right_high_ratio[0]*h, right_high_ratio[1]*h)
tl = (0, left_high_factor)
bl = (0, left_high_factor+left_low_factor)
tr = (w, right_high_factor)
br = (w, right_high_factor+right_low_factor)
contour = np.array([tl, tr, br, bl], dtype=np.int32)
mask = np.zeros(img.shape, dtype=img.dtype)
cv2.fillPoly(mask,[contour],(255,255,255))
mask = cv2.cvtColor(mask, cv2.COLOR_BGR2GRAY)
inverted_mask = cv2.bitwise_not(mask)
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
hsv[..., 2] = cv2.multiply(hsv[..., 2], high_bright_factor)
high_brightness = cv2.cvtColor(hsv, cv2.COLOR_HSV2BGR)
hsv[..., 2] = cv2.multiply(hsv[..., 2], low_bright_factor)
low_brightness = cv2.cvtColor(hsv, cv2.COLOR_HSV2BGR)
for i in range(3):
img[:, :, i] = img[:, :, i] *(mask/255) + high_brightness[:, :, i] *(1-mask/255)
img[:, :, i] = img[:, :, i] *(inverted_mask/255) + low_brightness[:, :, i] *(1-inverted_mask/255)
img = np.clip(img/255.0, 0, 1).astype(np.float32)
img = imagelib.warp_by_params (warp_params, img, warp, transform, can_flip=True, border_replicate=border_replicate) img = imagelib.warp_by_params (warp_params, img, warp, transform, can_flip=True, border_replicate=border_replicate)
img = np.clip(img.astype(np.float32), 0, 1) img = np.clip(img.astype(np.float32), 0, 1)