-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathvisualize.py
137 lines (130 loc) · 5.62 KB
/
visualize.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
import numpy as np
from PIL import Image
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
from util import *
import os
from tqdm import tqdm
import json
from model import SimpleNet,NewNet
#相机坐标架可视化
def visualize_frame():
fig = plt.figure()
ax = fig.add_subplot(projection = '3d')
ax.set_xlabel('X label') # 画出坐标轴
ax.set_ylabel('Y label')
ax.set_zlabel('Z label')
for j in range(3):
phi = np.linspace(0,np.pi,9)
theta = -np.pi/6/2*j
c2ws = get_c2w(phi=phi,theta=theta)
for i in range(9):
c2w = c2ws[i]
xs = c2w@(np.array([[j/5,0,0,1] for j in range(2)])[:,:,None])
ys = c2w@(np.array([[0,j/5,0,1] for j in range(2)])[:,:,None])
zs = c2w@(np.array([[0,0,j/5,1] for j in range(2)])[:,:,None])
assert(xs[0,3,0] == 1)
# assert(np.dot(xs[1,:,0] - xs[0,:,0],ys[1,:,0] - ys[0,:,0]) == 0 and
# np.dot(xs[1,:,0] - xs[0,:,0],zs[1,:,0] - zs[0,:,0]) == 0)
ax.plot(xs[:,0,0],xs[:,1,0],xs[:,2,0],color="red")
ax.plot(ys[:,0,0],ys[:,1,0],ys[:,2,0],color="blue")
ax.plot(zs[:,0,0],zs[:,1,0],zs[:,2,0],color="green")
plt.axis("equal")
plt.show()
#相机光线与采样点可视化
def visualize_ray():
fig = plt.figure()
ax = fig.add_subplot(projection = '3d')
ax.set_xlabel('X label') # 画出坐标轴
ax.set_ylabel('Y label')
ax.set_zlabel('Z label')
resolution = (5,5)
intrinsics = (r,r,r)
for j in range(1):
phi = np.linspace(0,np.pi,4)
theta = -np.pi/6/2*j
c2ws = get_c2w(phi=phi,theta=theta)
for i in range(4):
c2w = c2ws[i:i+1]
xs = c2w@(np.array([[j/5,0,0,1] for j in range(2)])[:,:,None])
ys = c2w@(np.array([[0,j/5,0,1] for j in range(2)])[:,:,None])
zs = c2w@(np.array([[0,0,j/5,1] for j in range(2)])[:,:,None])
assert(xs[0,3,0] == 1)
ax.plot(xs[:,0,0],xs[:,1,0],xs[:,2,0],color="red")
ax.plot(ys[:,0,0],ys[:,1,0],ys[:,2,0],color="blue")
ax.plot(zs[:,0,0],zs[:,1,0],zs[:,2,0],color="green")
H,W,focal = intrinsics
# for a in range(resolution[0]):
# for b in range(resolution[1]):
# h,w = a/(resolution[0]-1)*H, b/(resolution[1]-1)*W
# direction,origin = get_ray(c2w,h,w,intrinsics) #n*3
# samples = uni_sample(0.5*r, 1.5*r)*direction[:,None,:] #(sample_num,1) * (n,1,3) = n*sample_num*3
# samples = samples + origin[:,None,:] #n*sample_num*3
# ax.scatter(samples[0][::3,0],samples[0][::3,1],samples[0][::3,2],s=5)
# dot = (origin + direction*0.5)[0]
# origin = origin[0]
# ax.plot([origin[0],dot[0]],[origin[1],dot[1]],[origin[2],dot[2]])
plt.axis("equal")
plt.show()
#场景渲染为图片
def visualize_scene(root,ckp):
view_num = 8
intrinsics = (r,r,r)
print(r)
resolution = (128,128)
net = NewNet("cuda").to("cuda")
# net.load_state_dict(torch.load("gauss_new_net_3000" + ".pth"))
net.load_state_dict(torch.load(root+'/'+ckp))
net.eval()
theta = -np.pi/12
phi_list = np.linspace(0,2*np.pi,view_num,endpoint=False)
print(phi_list)
c2w = get_c2w(phi=phi_list,theta=theta)
with torch.no_grad():
net.eval()
color_img,trans_img,_ = render_image(net,c2w,intrinsics,resolution,1,"cuda")
# print(color_img,trans_img)
color_img = color_img.permute(0,3,1,2)
color_img = F.interpolate(color_img, size=(224,224))
color_img = color_img.permute(0,2,3,1)
file_name = root+'/'+ckp[:-4]+'_img'
if not os.path.isdir(file_name):
os.mkdir(file_name)
for i in range(view_num):
# print(i,color_img[i])
img_pil = Image.fromarray(np.uint8(color_img[i].cpu().numpy()*255))
img_pil.save(file_name+'/'+ckp[:-4]+f"_{i}.jpg")
#可视化NeRF数据集的相机位置（在本项目中没有使用）
def visualize_synthetic():
with open('./data/nerf_synthetic/lego/transforms_train.json') as f:
data = json.load(f)
frames = data['frames']
view_num = len(frames)
transfrom_matrixs = []
for i in range(view_num):
transfrom_matrixs.append(frames[i]['transform_matrix'])
transfrom_matrixs = np.array(transfrom_matrixs)
print(transfrom_matrixs.shape) #(100,4,4)
fig = plt.figure()
ax = fig.add_subplot(projection = '3d')
ax.set_xlabel('X label') # 画出坐标轴
ax.set_ylabel('Y label')
ax.set_zlabel('Z label')
for i in range(9):
c2w = transfrom_matrixs[i]
xs = c2w@(np.array([[j/5,0,0,1] for j in range(2)])[:,:,None])
ys = c2w@(np.array([[0,j/5,0,1] for j in range(2)])[:,:,None])
zs = c2w@(np.array([[0,0,j/5,1] for j in range(2)])[:,:,None])
assert(xs[0,3,0] == 1)
# assert(np.dot(xs[1,:,0] - xs[0,:,0],ys[1,:,0] - ys[0,:,0]) == 0 and
# np.dot(xs[1,:,0] - xs[0,:,0],zs[1,:,0] - zs[0,:,0]) == 0)
ax.plot(xs[:,0,0],xs[:,1,0],xs[:,2,0],color="red")
ax.plot(ys[:,0,0],ys[:,1,0],ys[:,2,0],color="blue")
ax.plot(zs[:,0,0],zs[:,1,0],zs[:,2,0],color="green")
plt.axis("equal")
plt.show()
if __name__=="__main__":
visualize_scene('results/gauss_4sample_range_test','gauss_4sample_range_test_2000.pth')
# visualize_frame()
# visualize_ray()
# visualize_synthetic()