Tobias Bertel, Yusuke Tomoto, Srinivas Rao, Rodrigo Ortiz-Cayon, Stefan Holzer and Christian Richardt
"Deferred Neural Rendering for View Extrapolation"
in ACM SIGGRAPH Asia 2020 posters
"Deferred Neural Rendering for View Extrapolation"
in ACM SIGGRAPH Asia 2020 posters
We capture an input video with a consumer camera, estimate camera poses, reconstruct a mesh and uv-map it.
We extend Deferred Neural Rendering [Thies et al. 2019] (blue) to enable smooth extrapolation of novel viewpoints (orange).
Abstract:
Image-based rendering methods that support visually pleasing specular surface reflections require accurate surface geometry and a large number of input images. Recent advances in neural scene
representations show excellent visual quality while requiring only imperfect mesh proxies or no surface-based proxies at all. While providing state-of-the-art visual quality, the inference time of learned models is usually too slow for interactive applications. While using a casually captured circular video sweep as input, we extend Deferred Neural Rendering to extrapolate smooth viewpoints around specular objects like a car.
representations show excellent visual quality while requiring only imperfect mesh proxies or no surface-based proxies at all. While providing state-of-the-art visual quality, the inference time of learned models is usually too slow for interactive applications. While using a casually captured circular video sweep as input, we extend Deferred Neural Rendering to extrapolate smooth viewpoints around specular objects like a car.
Submission video