-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
295 lines (260 loc) · 12.5 KB
/
index.html
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
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description" content="InverseShapeSVBRDF.">
<meta name="keywords" content="inverse rendering, SVBRDF, reconstruction">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Unified Shape and SVBRDF Recovery using Differentiable Monte Carlo Rendering</title>
<script>
window.dataLayer = window.dataLayer || [];
function gtag() {
dataLayer.push(arguments);
}
gtag('js', new Date());
gtag('config', 'G-PYVRSFMDRL');
</script>
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
<!-- Import maps polyfill -->
<!-- Remove this when import maps will be widely supported -->
<script async src="https://unpkg.com/es-module-shims@1.3.6/dist/es-module-shims.js"></script>
<script type="importmap">
{
"imports": {
"three": "https://unpkg.com/three/build/three.module.js",
"three/addons/": "https://unpkg.com/three/examples/jsm/"
}
}
</script>
<!-- <script src="https://threejs.org/build/three.js"></script> -->
<!-- <script type="module" src='https://threejs.org/examples/jsm/controls/OrbitControls.js'></script> -->
<!-- <script type="module" src='https://threejs.org/examples/jsm/loaders/OBJLoader.js'></script> -->
<script type="module" src="./static/js/mesh.js"></script>
</head>
<body>
<section class="hero">
<div class="hero-body">
<div class="container">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">Unified Shape and SVBRDF Recovery using<br> Differentiable Monte Carlo Rendering</h1>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a href="https://www.cs.cornell.edu/~fujun/">Fujun Luan</a><sup>1,3</sup>,</span>
<span class="author-block">
<a href="https://www.shuangz.com/">Shuang Zhao</a><sup>2</sup>,</span>
<span class="author-block">
<a href="https://www.cs.cornell.edu/~kb/">Kavita Bala</a><sup>1</sup>,
</span>
<span class="author-block">
<a href="http://flycooler.com/">Zhao Dong</a><sup>3</sup>
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup>Cornell University </span>
<span class="author-block"><sup>2</sup>University of California, Irvine </span>
<span class="author-block"><sup>3</sup>Facebook Reality Labs </span>
</div>
<h1 style="font-size:24px;font-weight:bold">EGSR 2021</h1>
<div class="column has-text-centered">
<div class="publication-links">
<span class="link-block">
<a href="https://www.cs.cornell.edu/~fujun/files/egsr2021/paper.pdf"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="link-block">
<a href="https://www.cs.cornell.edu/~fujun/files/egsr2021/supp.pdf"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Supplementary</span>
</a>
</span>
<span class="link-block">
<a href="https://arxiv.org/abs/2103.15208"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span>arXiv</span>
</a>
</span>
<span class="link-block">
<a href="https://psdr-cuda.readthedocs.io/en/latest/"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero teaser">
<div class="hero-body">
<div class="columns is-centered">
<div class="column is-8">
<span id="teaser_img" ><img id="teaser" width="100%" src="static/imgs/teaser.png"/></span>
<h2 class="subtitle has-text-centered">
Our approach takes multi-view photos of an object as input and uses Monte Carlo differentiable rendering to jointly reconstruct a set of PBR textures and a triangle mesh, which can easily be re-used in traditional computer graphics pipeline such as a path tracer or AR/VR environment.
</h2>
<img id="teaser2" width="100%" src="static/imgs/3d_modeling.jpg"/>
<h2 class="subtitle has-text-centered">
Using our reconstructed 3D models in Blender (Cycle rendering).
</h2>
<img id="teaser2" width="100%" src="static/imgs/ar.png"/>
<h2 class="subtitle has-text-centered">
Using our reconstructed 3D models in AR-based object insertion.
</h2>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-two-thirds">
<h2 class="title is-2">Abstract</h2>
<div class="content has-text-justified">
<p>
Reconstructing the shape and appearance of real-world objects using measured 2D images has been a long-standing problem in computer vision.
In this paper, we introduce a new analysis-by-synthesis technique capable of producing high-quality reconstructions through robust coarse-to-fine optimization and physics-based differentiable rendering.<br><br>
Unlike most previous methods that handle geometry and reflectance largely separately, our method unifies the optimization of both by leveraging image gradients with respect to both object reflectance <em>and</em> geometry.
To obtain physically accurate gradient estimates, we develop a new GPU-based Monte Carlo differentiable renderer leveraging recent advances in differentiable rendering theory to offer unbiased gradients while enjoying better performance than existing tools like PyTorch3D and redner.
To further improve robustness, we utilize several shape and material priors as well as a coarse-to-fine optimization strategy to reconstruct geometry.
We demonstrate that our technique can produce reconstructions with higher quality than previous methods such as COLMAP and Kinect Fusion.
</p>
</div>
</div>
</div>
</div>
<!--/ Abstract. -->
</section>
<section class="hero results">
<div class="hero-body">
<div class="columns is-centered">
<div class="column is-8">
<video id="v0" width="100%" autoplay="" loop="" muted="" controls="">
<source src="static/videos/table.mp4" type="video/mp4">
</video>
<video id="v1" width="100%" autoplay="" loop="" muted="" controls="">
<source src="static/videos/office.mp4" type="video/mp4">
</video>
<img id="result" width="100%" src="static/imgs/result.png"/>
<img id="more_res1" width="100%" src="static/imgs/more_res1.png"/>
<img id="more_res_bell" width="100%" src="static/imgs/more_res_bell.png"/>
<h2 class="subtitle has-text-centered">
More results of our method. For each object, typically 50~100 views are captured on the hemisphere.
</h2>
</div>
</div>
</div>
</section>
<section class="hero ipad">
<div class="hero-body">
<div class="columns is-centered">
<div class="column is-8">
<h2 class="subtitle has-text-centered">
With <a href="https://www.apple.com/newsroom/2020/03/apple-unveils-new-ipad-pro-with-lidar-scanner-and-trackpad-support-in-ipados/">iPad Pro LiDAR</a> depth sensor, we can get a quick 3D scan of an object (try the WebGL object on the right, obtained using <a href="https://scaniverse.com/">Scaniverse</a> app).
</h2>
<div style="width: 49.9%; float:left">
<img width="100%" src="static/imgs/ipad_lidar.jpg"/>
</div>
<div style="width: 49.9%; float:right" id="ipad">
</div>
<div style="width: 99.8%;">
<img width="100%" src="static/imgs/more_res_rabbit.png"/>
</div>
<h2 class="subtitle has-text-centered">
We can then jointly refine its geometry and appearance using our pipeline and re-use the reconstructed 3D model in AR/VR.
</h2>
</div>
</div>
</div>
</section>
<section class="hero azure">
<div class="hero-body">
<div class="columns is-centered">
<div class="column is-8">
<h2 class="subtitle has-text-centered">
With other low-cost depth sensors such as <a href="https://azure.microsoft.com/en-us/services/kinect-dk/">Microsoft Azure Kinect</a> or <a href="https://structure.io/structure-core">Structure Core</a>, we can also get a quick 3D scan of an object (try the WebGL object on the right, obtained using <a href="https://skanect.occipital.com/">Skanect Pro</a>).
</h2>
<div style="width: 49.9%; float:left">
<img width="100%" src="static/imgs/azure.jpg"/>
</div>
<div style="width: 49.9%; float:right" id="azure">
</div>
<div style="width: 99.8%;">
<img width="100%" src="static/imgs/more_res_turtle.png"/>
</div>
<h2 class="subtitle has-text-centered">
Again, we can refine and re-use the reconstructed 3D model in AR/VR.
</h2>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container">
<!-- Ack. -->
<div class="columns is-centered has-text-centered">
<div class="column is-two-thirds">
<h2 class="title is-2">Acknowledgements</h2>
<div class="content has-text-justified">
<p>
We thank Chenglei Wu, Yujia Chen, Christoph lassner, Sai Bi, Zhengqin Li, Giljoo Nam, Yue Dong, Hongzhi Wu, Zhongshi Jiang as well as the anonymous reviewers for their valuable discussions. We thank the digital artist James Warren from Facebook Reality Labs for modeling and rendering the two table scenes, and Inseung Hwang from KAIST for making comparisons with [Nam et al. 2018]. This work was supported in part by NSF grants 1900783 and 1900927.
</p>
</div>
</div>
</div>
</div>
<!--/ Ack. -->
</section>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code>
@article{luan2021unified,
author = {Luan, Fujun and Zhao, Shuang and Bala, Kavita and Dong, Zhao},
title = {Unified Shape and SVBRDF Recovery using Differentiable Monte Carlo Rendering},
journal = {ArXiv},
year = {2021},
}
</code></pre>
</div>
</section>
<footer class="footer">
<div align="center" class="container">
<div class="columns is-centered">
<div class="content">
This website is borrowed from <a href="https://github.com/nerfies/nerfies.github.io">nerfies</a>.
</div>
</div>
</div>
</footer>
</body>
</html>