-
Notifications
You must be signed in to change notification settings - Fork 9
/
index.html
282 lines (260 loc) · 12.7 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
<!DOCTYPE html>
<html>
<head>
<!-- Google tag (gtag.js) -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-CS1E2LNG5X"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-CS1E2LNG5X');
</script>
<meta charset="utf-8">
<meta name="description" content="GameNGen">
<meta property="og:title" content="GameNGen" />
<meta property="og:description" content="Diffusion Models Are Real-Time Game Engines" />
<meta property="og:url" content="https://gamengen.github.io/" />
<meta property="og:image" content="static/images/embed_image.png" />
<meta property="og:image:width" content="905" />
<meta property="og:image:height" content="375" />
<meta name="twitter:title" content="GameNGen">
<meta name="twitter:description" content="Diffusion Models Are Real-Time Game Engines">
<meta name="twitter:image" content="static/images/embed_image.png">
<meta name="twitter:card" content="GameNGen">
<meta name="keywords" content="GameNGen">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>GameNGen</title>
<link rel="icon" type="image/x-icon" href="static/images/favicon.ico">
<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 src="https://documentcloud.adobe.com/view-sdk/main.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>
</head>
<body>
<section class="hero">
<div class="hero-body" style="padding-bottom: 1rem">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">Diffusion Models Are Real-Time Game Engines</h1>
<div class="is-size-4 publication-authors">
<div class="author-block">
<span>
<a href="https://www.linkedin.com/in/dani-valevski-a3b5936/" target="_blank">Dani
Valevski</a><sup>*</sup>
</span>
<div class="author-affiliation">Google Research</div>
</div>
<div class="author-block">
<span>
<a href="https://yanivle.github.io/" target="_blank">Yaniv Leviathan</a><sup>*</sup>
</span>
<div class="author-affiliation">Google Research</div>
</div>
<div class="author-block">
<span>
<a href="https://moabarar.github.io/" target="_blank">Moab Arar</a><sup>*†</sup>
</span>
<div class="author-affiliation">Tel Aviv University</div>
</div>
<div class="author-block">
<span>
<a href="https://shlomifruchter.github.io/" target="_blank">Shlomi Fruchter</a><sup>*</sup>
</span>
<div class="author-affiliation">Google DeepMind</div>
</div>
<span class="eql-cntrb"><small><br><sup>*</sup>Equal Contribution <sup>†</sup>Work done while at Google
Research</small></span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/pdf/2408.14837" target="_blank"
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://arxiv.org/abs/2408.14837" target="_blank"
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>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero is-small">
<div class="hero-body" style="padding-top: 0">
<div class="container">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths" style="max-width:960px">
<div class="publication-video">
<iframe
src="https://www.youtube.com/embed/O3616ZFGpqw?autoplay=1&mute=1&loop=1&showinfo=0&list=PL3ZfMho22LwDvJSEKVBiwxNsVEqUTUmhJ"
frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>
</div>
</div>
</div>
<div class="subtitle has-text-centered">
<span>Real-time</span> recordings of people playing the game <a
href="https://en.wikipedia.org/wiki/Doom_(1993_video_game)" target="_blank">DOOM</a>
<span><b>simulated entirely by the <i>GameNGen</i> neural model</b></span>.
</div>
</div>
</div>
</section>
<section class="section hero is-light">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="abstract content has-text-justified">
<p>
We present <i>GameNGen</i>, the first game engine powered entirely by a neural model
that enables real-time interaction with a complex environment over long trajectories at high quality.
<i>GameNGen</i> can interactively simulate the classic game DOOM at over 20 frames per second on a single TPU.
Next frame prediction achieves a PSNR of 29.4, comparable to lossy JPEG compression.
Human raters are only slightly better than random chance at distinguishing short clips of the game from clips of the simulation.
<i>GameNGen</i> is trained in two phases: (1) an RL-agent learns to play the game and the training sessions are recorded, and
(2) a diffusion model is trained to produce the next frame, conditioned on the sequence of past frames and actions.
Conditioning augmentations enable stable auto-regressive generation over long trajectories.
</p>
</div>
</div>
</div>
</div>
</section>
<section class="hero is-small">
<div class="hero-body">
<div class="container is-centered has-text-centered">
<h2 class="title is-3">Full Gameplay Videos</h2>
<div id="results-carousel" class="carousel results-carousel">
<div class="item item-video1" style="max-width: 100%;">
<video poster id="video1" muted autoplay controls loop playsinline height="100%"
src="static/videos/e1m1_t.mp4">
</video>
</div>
<div class="item item-video2">
<video poster id="video2" muted autoplay controls loop playsinline height="100%"
src="static/videos/e1m3.mp4">
</video>
</div>
<div class="item item-video3">
<video poster id="video3" muted autoplay controls loop playsinline height="100%"
src="static/videos/e1m5.mp4" type="video/mp4">
</video>
</div>
<div class="item item-video4">
<video poster id="video4" muted autoplay controls loop playsinline height="100%"
src="static/videos/e1m9.mp4" type="video/mp4">
</video>
</div>
<div class="item item-video5">
<video poster id="video5" muted autoplay controls loop playsinline height="100%"
src="static/videos/e2m2.mp4" type="video/mp4">
</video>
</div>
</div>
</div>
</div>
</section>
<section class="section hero is-small">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Architecture</h2>
<div class="content has-text-justified">
<p></p>
<img src="static/images/Architecture_08_27.png" alt="Architecture overview" />
<p><b>Data Collection via Agent Play:</b>
Since we cannot collect human gameplay at scale, as a first stage we train an automatic RL-agent to play
the game,
persisting it's training episodes of actions and observations, which become the training data for our
generative model.
</p>
<p><b>Training the Generative Diffusion Model:</b> We re-purpose a small diffusion model, <a
href="https://arxiv.org/abs/2112.10752" target="_blank">Stable Diffusion</a> v1.4,
and condition it on a sequence of previous actions and observations (frames). To mitigate
auto-regressive drift
during inference, we corrupt context frames by adding Gaussian noise to encoded frames during training.
This allows the network to correct information sampled in previous frames, and we found it to be
critical
for preserving visual stability over long time periods.
</p>
<p><b>Latent Decoder Fine-Tuning:</b> The pre-trained auto-encoder of <a
href="https://arxiv.org/abs/2112.10752" target="_blank">Stable Diffusion</a> v1.4, which compresses
8x8
pixel patches into 4 latent channels, results in meaningful artifacts when predicting game frames, which
affect
small details and particularly the bottom bar HUD. To leverage the pre-trained knowledge while improving
image quality, we train just the decoder of the latent auto-encoder using an MSE loss computed
against the target frame pixels.
</p>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code>@misc{valevski2024diffusionmodelsrealtimegame,
title={Diffusion Models Are Real-Time Game Engines},
author={Dani Valevski and Yaniv Leviathan and Moab Arar and Shlomi Fruchter},
year={2024},
eprint={2408.14837},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2408.14837},
}</code></pre>
</div>
</section>
<section class="section" id="Acknowledgements">
<div class="container is-max-desktop content">
<h2 class="title">Acknowledgements</h2>
We'd like to extend a huge thank you to Eyal Segalis, Eyal Molad, Matan Kalman, Nataniel Ruiz,
Amir Hertz, Matan Cohen, Yossi Matias, Yael Pritch, Danny Lumen, Valerie Nygaard, the Theta Labs and Google
Research teams, and our families for insightful feedback, ideas, suggestions, and support.
</div>
</section>
<footer class="footer">
<div class="container">
<div class="columns is-centered">
<div class="column is-8">
<div class="content">
<p>
This page was built using the <a href="https://github.com/eliahuhorwitz/Academic-project-page-template"
target="_blank">Academic Project Page Template</a> which was adopted from the <a
href="https://nerfies.github.io" target="_blank">Nerfies</a> project page.
</p>
</div>
</div>
</div>
</div>
</footer>
<!-- Statcounter tracking code -->
<!-- You can add a tracker to track page visits by creating an account at statcounter.com -->
<!-- End of Statcounter Code -->
</body>
</html>