-
Notifications
You must be signed in to change notification settings - Fork 6
/
render.py
219 lines (185 loc) · 9.17 KB
/
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
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
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
#
# Copyright (C) 2023, Inria
# GRAPHDECO research group, https://team.inria.fr/graphdeco
# All rights reserved.
#
# This software is free for non-commercial, research and evaluation use
# under the terms of the LICENSE.md file.
#
# For inquiries contact george.drettakis@inria.fr
#
import imageio
import numpy as np
import torch
from scene import Scene
import os
from tqdm import tqdm
from os import makedirs
from gaussian_renderer import render
import torchvision
from utils.general_utils import safe_state
from argparse import ArgumentParser
from arguments import ModelParams, PipelineParams, get_combined_args, ModelHiddenParams
from gaussian_renderer import GaussianModel
from time import time
import open3d as o3d
from utils.graphics_utils import fov2focal
import cv2
import os
cpu_list = [0]
os.sched_setaffinity(0, cpu_list)
to8b = lambda x : (255*np.clip(x.cpu().numpy(),0,1)).astype(np.uint8)
def render_set(model_path, name, iteration, views, gaussians, pipeline, background, no_fine, render_test=False, reconstruct=False):
render_path = os.path.join(model_path, name, "ours_{}".format(iteration), "renders")
depth_path = os.path.join(model_path, name, "ours_{}".format(iteration), "depth")
gts_path = os.path.join(model_path, name, "ours_{}".format(iteration), "gt")
gtdepth_path = os.path.join(model_path, name, "ours_{}".format(iteration), "gt_depth")
masks_path = os.path.join(model_path, name, "ours_{}".format(iteration), "masks")
makedirs(render_path, exist_ok=True)
makedirs(depth_path, exist_ok=True)
makedirs(gts_path, exist_ok=True)
makedirs(gtdepth_path, exist_ok=True)
makedirs(masks_path, exist_ok=True)
render_images = []
render_depths = []
gt_list = []
gt_depths = []
mask_list = []
for idx, view in enumerate(tqdm(views, desc="Rendering progress")):
stage = 'coarse' if no_fine else 'fine'
rendering = render(view, gaussians, pipeline, background, stage=stage)
render_depths.append(rendering["depth"].cpu())
render_images.append(rendering["render"].cpu())
if name in ["train", "test", "video"]:
gt = view.original_image[0:3, :, :]
gt_list.append(gt)
mask = view.mask
mask_list.append(mask)
gt_depth = view.original_depth
gt_depths.append(gt_depth)
if render_test:
test_times = 50
for i in range(test_times):
for idx, view in enumerate(tqdm(views, desc="Rendering progress")):
if idx == 0 and i == 0:
time1 = time()
stage = 'coarse' if no_fine else 'fine'
rendering = render(view, gaussians, pipeline, background, stage=stage)
time2=time()
print("FPS:",(len(views)-1)*test_times/(time2-time1))
count = 0
print("writing training images.")
if len(gt_list) != 0:
for image in tqdm(gt_list):
torchvision.utils.save_image(image, os.path.join(gts_path, '{0:05d}'.format(count) + ".png"))
count+=1
count = 0
print("writing rendering images.")
if len(render_images) != 0:
for image in tqdm(render_images):
torchvision.utils.save_image(image, os.path.join(render_path, '{0:05d}'.format(count) + ".png"))
count +=1
count = 0
print("writing mask images.")
if len(mask_list) != 0:
for image in tqdm(mask_list):
image = image.float()
torchvision.utils.save_image(image, os.path.join(masks_path, '{0:05d}'.format(count) + ".png"))
count +=1
count = 0
print("writing rendered depth images.")
if len(render_depths) != 0:
for image in tqdm(render_depths):
image = np.clip(image.cpu().squeeze().numpy().astype(np.uint8), 0, 255)
cv2.imwrite(os.path.join(depth_path, '{0:05d}'.format(count) + ".png"), image)
count += 1
count = 0
print("writing gt depth images.")
if len(gt_depths) != 0:
for image in tqdm(gt_depths):
image = image.cpu().squeeze().numpy().astype(np.uint8)
cv2.imwrite(os.path.join(gtdepth_path, '{0:05d}'.format(count) + ".png"), image)
count += 1
render_array = torch.stack(render_images, dim=0).permute(0, 2, 3, 1)
render_array = (render_array*255).clip(0, 255).cpu().numpy().astype(np.uint8)
imageio.mimwrite(os.path.join(model_path, name, "ours_{}".format(iteration), 'ours_video.mp4'), render_array, fps=30, quality=8)
gt_array = torch.stack(gt_list, dim=0).permute(0, 2, 3, 1)
gt_array = (gt_array*255).clip(0, 255).cpu().numpy().astype(np.uint8)
imageio.mimwrite(os.path.join(model_path, name, "ours_{}".format(iteration), 'gt_video.mp4'), gt_array, fps=30, quality=8)
FoVy, FoVx, height, width = view.FoVy, view.FoVx, view.image_height, view.image_width
focal_y, focal_x = fov2focal(FoVy, height), fov2focal(FoVx, width)
camera_parameters = (focal_x, focal_y, width, height)
if reconstruct:
reconstruct_point_cloud(render_images, mask_list, render_depths, camera_parameters, name)
def render_sets(dataset : ModelParams, hyperparam, iteration : int, pipeline : PipelineParams, skip_train : bool, skip_test : bool, skip_video: bool, reconstruct: bool):
cpu_list = list(range(cpu_count))[1:2]
psutil.Process().cpu_affinity(cpu_list)
with torch.no_grad():
gaussians = GaussianModel(dataset.sh_degree, hyperparam)
scene = Scene(dataset, gaussians, load_iteration=iteration, shuffle=False, load_coarse=dataset.no_fine)
bg_color = [1,1,1] if dataset.white_background else [0, 0, 0]
background = torch.tensor(bg_color, dtype=torch.float32, device="cuda")
if not skip_train:
render_set(dataset.model_path, "train", scene.loaded_iter, scene.getTrainCameras(), gaussians, pipeline, background, dataset.no_fine, reconstruct=False)
if not skip_test:
render_set(dataset.model_path, "test", scene.loaded_iter, scene.getTestCameras(), gaussians, pipeline, background, dataset.no_fine, reconstruct=reconstruct)
if not skip_video:
render_set(dataset.model_path,"video",scene.loaded_iter, scene.getVideoCameras(),gaussians,pipeline,background, dataset.no_fine, render_test=True, reconstruct=False)
def reconstruct_point_cloud(images, masks, depths, camera_parameters, name):
import cv2
import copy
output_frame_folder = os.path.join("reconstruct", name)
os.makedirs(output_frame_folder, exist_ok=True)
frames = np.arange(len(images))
# frames = [0]
focal_x, focal_y, width, height = camera_parameters
for i_frame in frames:
rgb_tensor = images[i_frame]
rgb_np = rgb_tensor.mul(255).add_(0.5).clamp_(0, 255).permute(1, 2, 0).contiguous().to("cpu").numpy()
depth_np = depths[i_frame].cpu().numpy()
depth_np = depth_np.squeeze(0)
mask = masks[i_frame]
mask = mask.squeeze(0).cpu().numpy()
rgb_new = copy.deepcopy(rgb_np)
depth_smoother = (128, 64, 64)
depth_np = cv2.bilateralFilter(depth_np, depth_smoother[0], depth_smoother[1], depth_smoother[2])
close_depth = np.percentile(depth_np[depth_np!=0], 5)
inf_depth = np.percentile(depth_np, 95)
depth_np = np.clip(depth_np, close_depth, inf_depth)
rgb_im = o3d.geometry.Image(rgb_new.astype(np.uint8))
depth_im = o3d.geometry.Image(depth_np)
rgbd_image = o3d.geometry.RGBDImage.create_from_color_and_depth(rgb_im, depth_im, convert_rgb_to_intensity=False)
pcd = o3d.geometry.PointCloud.create_from_rgbd_image(
rgbd_image,
o3d.camera.PinholeCameraIntrinsic(width, height, focal_x, focal_y, width / 2, width / 2),
project_valid_depth_only=True
)
o3d.io.write_point_cloud(os.path.join(output_frame_folder, 'frame_{}.ply'.format(i_frame)), pcd)
if __name__ == "__main__":
import psutil
cpu_count = psutil.cpu_count()
print(cpu_count)
cpu_list = list(range(cpu_count))[1:2]
psutil.Process().cpu_affinity(cpu_list)
# Set up command line argument parser
parser = ArgumentParser(description="Testing script parameters")
model = ModelParams(parser, sentinel=True)
pipeline = PipelineParams(parser)
hyperparam = ModelHiddenParams(parser)
parser.add_argument("--iteration", default=-1, type=int)
parser.add_argument("--skip_train", action="store_true")
parser.add_argument("--skip_test", action="store_true")
parser.add_argument("--quiet", action="store_true")
parser.add_argument("--skip_video", action="store_true")
parser.add_argument("--configs", type=str)
parser.add_argument("--reconstruct", action="store_true")
args = get_combined_args(parser)
print("Rendering ", args.model_path)
if args.configs:
import mmcv
from utils.params_utils import merge_hparams
config = mmcv.Config.fromfile(args.configs)
args = merge_hparams(args, config)
# Initialize system state (RNG)
safe_state(args.quiet)
render_sets(model.extract(args), hyperparam.extract(args), args.iteration, pipeline.extract(args), args.skip_train, args.skip_test, args.skip_video, args.reconstruct)