-
Notifications
You must be signed in to change notification settings - Fork 5
/
search_and_render.py
125 lines (94 loc) · 5.66 KB
/
search_and_render.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
import math, tqdm
from Line_Tools import *
from simulate_RGB import *
from drawpatch import *
def Search_Stroke(anchors, density, img_gray, input_hsv, gradient, RATIO,
angle, angle_hatch, min_width, min_length, max_width, max_length, threshold_hsv):
(h0,w0) = img_gray.shape
patch_sequence = []
for i in tqdm.tqdm(range(anchors.shape[0])):
patch_temp = {}
# get coordinate
(x,y) = anchors[i]
y = h0 - y # y is Upside-down ！！！
y = np.clip(y,0,h0-1)
x = np.clip(x,0,w0-1)
x = int(round(x)) # for savety
y = int(round(y)) # for savety
patch_temp['coordinate'] = [y,x]
patch_temp['angle_hatch'] = angle_hatch[y,x]
patch_temp['path_hatch'], patch_temp['w1'], patch_temp['w2'] = \
FindPath_V7(img_gray, input_hsv, angle_hatch, y, x, step_length=1, min_length=min_width,max_length=max_width, threshold_hsv= threshold_hsv)
patch_temp['angle_ETF'] = angle[y,x]
patch_temp['path_ETF'], patch_temp['l1'] , patch_temp['l2'] = \
FindPath_V7(img_gray, input_hsv, angle, y, x, step_length=1, min_length=min_length,max_length=max_length, threshold_hsv= threshold_hsv)
patch_temp['grayscale'] = img_gray[y,x]
patch_temp['hsv'] = input_hsv[y,x]
patch_temp['gradient'] = gradient[y,x]
patch_temp['density'] = density[y,x]
max_local_width = np.sqrt(1.0/patch_temp['density'])
patch_temp['w1'] = min(patch_temp['w1'], max_local_width)
patch_temp['w1'] = max(patch_temp['w1'], min_width)
patch_temp['w2'] = min(patch_temp['w2'], max_local_width)
patch_temp['w2'] = max(patch_temp['w2'], min_width)
patch_temp['l1'] = min(patch_temp['l1'], RATIO*max_local_width)
patch_temp['l1'] = max(patch_temp['l1'], min_length)
patch_temp['l2'] = min(patch_temp['l2'], RATIO*max_local_width)
patch_temp['l2'] = max(patch_temp['l2'], min_length)
patch_temp['importance'] = (patch_temp['w1']+patch_temp['w2']+1)*(patch_temp['l1']+patch_temp['l2']+1) + np.random.randn()/100
patch_sequence.append(patch_temp)
return patch_sequence
def Render_Stroke(brush, patch_sequence, img_gray, output_path, max_length, SSAA, BORDERCOPY, FREQ, save=True):
(h0,w0) = img_gray.shape
Canvas = Gassian_HSV((h0*SSAA+2*max_length*SSAA,w0*SSAA+2*max_length*SSAA,3)) # padding max_length
Mask = np.zeros((h0*SSAA+2*max_length*SSAA,w0*SSAA+2*max_length*SSAA))
for step in tqdm.tqdm(range(len(patch_sequence))):
patch_temp = patch_sequence[step]
w1 = patch_temp['w1']
w2 = patch_temp['w2']
l1 = patch_temp['l1']
l2 = patch_temp['l2']
length = int(round((l1+1+l2)*SSAA))
width = int(round((w1+1+w2)*SSAA))
angle_ETF = patch_temp['angle_ETF']
angle_Hatch = angle_Hatch = patch_temp['angle_hatch']
# get patch central coordinate
central = patch_temp['coordinate']
central[1]-= (l1-l2)/2*math.cos(angle_ETF/180*math.pi) # x
central[0]+= (l1-l2)/2*math.sin(angle_ETF/180*math.pi) # y
central[1]-= (w1-w2)/2*math.cos(angle_Hatch/180*math.pi) # x
central[0]+= (w1-w2)/2*math.sin(angle_Hatch/180*math.pi) # y
Y, X = int(round(central[0]*SSAA)), int(round(central[1]*SSAA))
Y += max_length*SSAA # Canvas edge is padded
X += max_length*SSAA # Canvas edge is padded
hsv = patch_temp['hsv']
stroke, mask = GetBrush_HSV(width=width, length=length, mean=hsv, brush=brush)
rotate_stroke, _ = rotate_hsv(stroke, angle_ETF, pad_color=(int(hsv[0]),0,255))
rotate_stroke = np.clip(rotate_stroke,0,255)
(rotate_h, rotate_w, _) = rotate_stroke.shape
rotate_hh, rotate_ww = int(rotate_h/2), int(rotate_w/2)
mask, _ = rotate(mask, angle_ETF, pad_color=0)
mask[mask<1] = 0
# cv2.imshow('Canvas', np.uint8(rotate_stroke))
# cv2.waitKey(0)
Mask[Y-rotate_hh:Y-rotate_hh+rotate_h, X-rotate_ww:X-rotate_ww+rotate_w] += mask
temp = Canvas[Y-rotate_hh:Y-rotate_hh+rotate_h, X-rotate_ww:X-rotate_ww+rotate_w]
rotate_alpha = mask.astype("float32")
rotate_alpha = rotate_alpha[:,:,np.newaxis]
rotate_stroke = rotate_stroke.astype("float32")
# white = Gassian_HSV(rotate_stroke.shape) # padding
# white = np.uint8((1-rotate_alpha)*white + rotate_alpha*rotate_stroke)
# cv2.imwrite(output_path + "/stroke/{0:05d}.png".format(step), cv2.cvtColor(white, cv2.COLOR_HSV2BGR))
Canvas[Y-rotate_hh:Y-rotate_hh+rotate_h, X-rotate_ww:X-rotate_ww+rotate_w] = np.uint8((1-rotate_alpha)*temp + rotate_alpha*rotate_stroke)
if save and (step+1) % FREQ == 0:
result = Canvas[max_length*SSAA:-max_length*SSAA,max_length*SSAA:-max_length*SSAA]
cv2.imwrite(output_path + "/process/{0:05d}.png".format(step+1), cv2.cvtColor(result[BORDERCOPY*SSAA:-BORDERCOPY*SSAA,BORDERCOPY*SSAA:-BORDERCOPY*SSAA], cv2.COLOR_HSV2BGR))
# cv2.imshow('step', Canvas)
# cv2.waitKey(0)
if save and step == len(patch_sequence)-1 and step+1 % FREQ != 0: # 存最后一张
result = Canvas[max_length*SSAA:-max_length*SSAA,max_length*SSAA:-max_length*SSAA]
cv2.imwrite(output_path + "/process/{0:05d}.png".format(step+1), cv2.cvtColor(result[BORDERCOPY*SSAA:-BORDERCOPY*SSAA,BORDERCOPY*SSAA:-BORDERCOPY*SSAA], cv2.COLOR_HSV2BGR))
Mask[Mask>0] = 1
# cv2.imshow('mask', np.uint8(Mask*255))
# cv2.waitKey(0)
return Canvas, Mask # hsv