/
blender_mesh_generator.py
154 lines (135 loc) · 6.25 KB
/
blender_mesh_generator.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
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
import time
import bpy
import random
import numpy as np; np.random.seed(20200527)
import tensorflow as tf
from skimage import morphology
print(tf.__version__)
def delete_all_obj():
for obj in bpy.data.objects:
bpy.data.objects.remove(bpy.data.objects[obj.name])
def create_cube(scale=[1.0, 1.0, 1.0], location=[0.0, 0.0, 0.0], rotation=[0.0, 0.0, 0.0]):
bpy.ops.mesh.primitive_cube_add(size=1.0, calc_uvs=True
, enter_editmode=False, align='WORLD'
, location=location, rotation=rotation
)
bpy.context.object.scale = scale
def create_beam_1x1(val, beam_length, location, rotation):
"""
通常はただの 1x1 の梁を生成。ピクセル値が中途半端なとき、部材を確率的に弱くし、均等さを保つ。
（このヒューリスティックは3Dプリンタと相性が悪かったかもしれない。）
"""
if(val == 1.0):
create_cube(scale=[1.0, 1.0, beam_length], location=location, rotation=rotation)
else:
_r = random.randint(0, 3)
if(_r==0):#(0, 1, 2, 3)
create_cube(scale=[0.5, 1.0, beam_length]
, location=location+[-0.25, 0.0, 0.0], rotation=rotation)
create_cube(scale=[0.5, 0.5, beam_length]
, location=location+[+0.25, +0.25, 0.0], rotation=rotation)
elif(_r==1):
create_cube(scale=[0.5, 1.0, beam_length]
, location=location+[-0.25, 0.0, 0.0], rotation=rotation)
create_cube(scale=[0.5, 0.5, beam_length]
, location=location+[+0.25, -0.25, 0.0], rotation=rotation)
elif(_r==2):
create_cube(scale=[0.5, 1.0, beam_length]
, location=location+[+0.25, 0.0, 0.0], rotation=rotation)
create_cube(scale=[0.5, 0.5, beam_length]
, location=location+[-0.25, +0.25, 0.0], rotation=rotation)
elif(_r==3):
create_cube(scale=[0.5, 1.0, beam_length]
, location=location+[+0.25, 0.0, 0.0], rotation=rotation)
create_cube(scale=[0.5, 0.5, beam_length]
, location=location+[-0.25, -0.25, 0.0], rotation=rotation)
else:
raise ValueError("out of range")
class Generator:
def __init__(self):
self.NOISE_DIM = 128
self.FIXED_NOISE_FOR_PREDICT = np.random.normal(0, 1, (1, self.NOISE_DIM))
self.model = self.build_generator()
self.model.load_weights("C:/Users/.../Documents/Blender/model-I_x+75-24000.hdf5")
def build_generator(self):
z = z_in = tf.keras.layers.Input(shape=(self.NOISE_DIM, ), name="noise")
x = tf.keras.layers.Dense(1024)(z)
x = tf.keras.layers.LeakyReLU(alpha=0.2)(x)
x = tf.keras.layers.BatchNormalization(momentum=0.8)(x)
x = tf.keras.layers.Dense(7*7*64)(z)
x = tf.keras.layers.LeakyReLU(alpha=0.2)(x)
x = tf.keras.layers.BatchNormalization(momentum=0.8)(x)
x = tf.keras.layers.Reshape(target_shape=(7, 7, 64))(x)
x = tf.keras.layers.Conv2DTranspose(32, kernel_size=(5, 5)
, padding='same', strides=(2, 2), use_bias=False, activation=None)(x)
x = tf.keras.layers.BatchNormalization(momentum=0.8)(x)
x = tf.keras.layers.LeakyReLU(alpha=0.2)(x)
x = tf.keras.layers.Conv2DTranspose(1, kernel_size=(5, 5)
, padding='same', strides=(2, 2), use_bias=False, activation=None)(x)
img = tf.math.tanh(x)
y = tf.keras.layers.Lambda(lambda x: x, name="generated_image")(img) #
img = (img + 1.0)/2.0
I_x, I_y, I_r = tf.reduce_sum(img), tf.reduce_sum(img), tf.reduce_sum(img)
return tf.keras.Model(inputs=z_in, outputs=[y, I_x, I_y, I_r])
def generate_number(self):
batch, _, _, _ = self.model.predict(self.FIXED_NOISE_FOR_PREDICT)
number = (batch[0, :, :, 0] + 1.0)/ 2.0
return number
class CreateDanmen:
def __init__(self):
"""
Initialie constants.
"""
self.NUM_X_BOXES = 28
self.NUM_Y_BOXES = 28
self.NUM_BOXES = self.NUM_X_BOXES * self.NUM_Y_BOXES
self.BEAM_LENGTH = 60
self.SUPPORT_THICKNESS = 12
"""
Load MNIST.
"""
(X_train, _), (_, _) = tf.keras.datasets.mnist.load_data()
X_train = (X_train / 255.0)#.astype(np.int)
"""
0.75 < x to 1
0.25 < x < 0.75 = 0.5 (variable)
x < 0.25 to 0
"""
if(True):val = X_train[1].reshape((28, 28))
val = Generator().generate_number()
val[0.75 <= val] = 1.0
val[np.where(np.asarray(0.25 < val) & np.asarray(val < 0.75))] = 0.5
val[val <= 0.25] = 0.0
self.VAL = val
print(val)
def boxloop(self, val, beam_length, offset):
# All boxes
for x in range(self.NUM_X_BOXES):
"""
Initialize all primitive params.
"""
box_loc = np.empty(shape=[self.NUM_Y_BOXES, 3], dtype=np.float32) # Variable
box_rot = np.zeros(shape=[self.NUM_Y_BOXES, 3], dtype=np.float32) # No rotation
for y in range(self.NUM_Y_BOXES):
if(val[x, y] == 0): continue # pass 0 value
box_loc[y] = [offset[0]+x, offset[1]+y, beam_length/2.0]
create_beam_1x1(val=val[x, y]
, beam_length=beam_length
, location=box_loc[y], rotation=box_rot[y]) # Cube
def create_body(self):
self.boxloop(val=self.VAL, beam_length=self.BEAM_LENGTH, offset=(0, 0))
#val_blur = filters.gaussian(self.VAL, sigma=3, mode='nearest')
"""
画像を反転してピンを作る
"""
half_mask = np.concatenate([np.zeros(shape=[14, 28]), np.ones(shape=[14, 28])], axis=0)
support_half = (1 - np.ceil(self.VAL)) * half_mask
support_erosion = morphology.binary_erosion(support_half, morphology.diamond(2)).astype(np.float32)
for i in range(2):
self.boxloop(val=support_erosion, beam_length=self.SUPPORT_THICKNESS, offset=(14*i, 0))
if __name__ == "__main__":
delete_all_obj()
start_time = time.time()
createdanmen = CreateDanmen()
createdanmen.create_body()
print(f"SpentTime: {time.time() - start_time}[s]")