FastNeRF: High-Fidelity Neural Rendering at 200FPS [Extended]
9 376
19.1
Microsoft Research335 тыс
Следующее
Популярные
51 день – 49449:37
65 дней – 7864:30Опубликовано 1 ноября 2021, 13:34
Watch the condensed version: youtu.be/JS5H-Usiphg
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
9 дней – 8 6781:30
9 дней – 1557:44
10 дней – 3716:51Случайные видео
29 дней – 1 024 5741:00
305 дней – 1 097 9230:29
28.07.20 – 54 4181:25
27.09.15 – 4 1863:57
1 день – 104 8430:38
2 дня – 37 4580:27
3 дня – 9310:50
4 дня – 130 27147:55
4 дня – 4490:17
5 дней – 3896:54
5 дней – 1372:45
18 часов – 871:05
2 дня – 2 0390:31