FastNeRF: High-Fidelity Neural Rendering at 200FPS [Condensed]
9 236
15.4
Microsoft Research333 тыс
Следующее
27.04.21 – 5732:59:41Популярные
192 дня – 6243:07:51
343 дня – 1 3391:21Опубликовано 26 апреля 2021, 17:16
Watch the extended version: youtu.be/mi5b142WEmw
Recent work on Neural Radiance Fields (NeRF) showed how neural networks can be used to encode complex 3D environments that can be rendered photorealistically from novel viewpoints. Rendering these images is very computationally demanding and recent improvements are still a long way from enabling interactive rates, even on high-end hardware. Motivated by scenarios on mobile and mixed reality devices, we propose FastNeRF, the first NeRF-based system capable of rendering high fidelity photorealistic images at 200Hz on a high-end consumer GPU. The core of our method is a graphics-inspired factorization that allows for (i) compactly caching a deep radiance map at each position in space, (ii) efficiently querying that map using ray directions to estimate the pixel values in the rendered image. Extensive experiments show that the proposed method is 3000 times faster than the original NeRF algorithm and at least an order of magnitude faster than existing work on accelerating NeRF, while maintaining visual quality and extensibility.
Project website: microsoft.github.io/FastNeRF
Mixed Reality & AI Lab: microsoft.com/en-us/research/l...
Recent work on Neural Radiance Fields (NeRF) showed how neural networks can be used to encode complex 3D environments that can be rendered photorealistically from novel viewpoints. Rendering these images is very computationally demanding and recent improvements are still a long way from enabling interactive rates, even on high-end hardware. Motivated by scenarios on mobile and mixed reality devices, we propose FastNeRF, the first NeRF-based system capable of rendering high fidelity photorealistic images at 200Hz on a high-end consumer GPU. The core of our method is a graphics-inspired factorization that allows for (i) compactly caching a deep radiance map at each position in space, (ii) efficiently querying that map using ray directions to estimate the pixel values in the rendered image. Extensive experiments show that the proposed method is 3000 times faster than the original NeRF algorithm and at least an order of magnitude faster than existing work on accelerating NeRF, while maintaining visual quality and extensibility.
Project website: microsoft.github.io/FastNeRF
Mixed Reality & AI Lab: microsoft.com/en-us/research/l...
Свежие видео
14 дней – 181 6953:01
15 дней – 3 59110:49
15 дней – 22 09433:15
19 дней – 2 5370:19
20 дней – 20 56814:29
21 день – 27 3583:25Случайные видео
65 дней – 26 8210:57
66 дней – 2640:19
150 дней – 6 2410:22
299 дней – 367 0230:40
16.01.23 – 6 6530:34
04.01.23 – 3 108 2501:00
3 дня – 1 5010:21
8 дней – 2511:52
8 дней – 32 6628:12
9 дней – 10 009 32812:03
15 дней – 92 2279:51
16 дней – 9111:00
20 дней – 4606:37
6 дней – 1921:09
6 дней – 4 049 26720:29