HybridNeRF: Efficient Neural Rendering via Adaptive Volumetric Surfaces
This addresses rendering speed issues in neural rendering for applications like virtual reality, though it is incremental as it builds on existing surface and volumetric methods.
The paper tackled the problem of slow rendering in neural radiance fields by proposing HybridNeRF, which combines surface and volumetric representations to improve efficiency; it achieved 15-30% lower error rates and real-time framerates of at least 36 FPS at 2Kx2K resolution.
Neural radiance fields provide state-of-the-art view synthesis quality but tend to be slow to render. One reason is that they make use of volume rendering, thus requiring many samples (and model queries) per ray at render time. Although this representation is flexible and easy to optimize, most real-world objects can be modeled more efficiently with surfaces instead of volumes, requiring far fewer samples per ray. This observation has spurred considerable progress in surface representations such as signed distance functions, but these may struggle to model semi-opaque and thin structures. We propose a method, HybridNeRF, that leverages the strengths of both representations by rendering most objects as surfaces while modeling the (typically) small fraction of challenging regions volumetrically. We evaluate HybridNeRF against the challenging Eyeful Tower dataset along with other commonly used view synthesis datasets. When comparing to state-of-the-art baselines, including recent rasterization-based approaches, we improve error rates by 15-30% while achieving real-time framerates (at least 36 FPS) for virtual-reality resolutions (2Kx2K).