Sampling for View Synthesis: From Local Light Field Fusion to Neural Radiance Fields and Beyond

TL;DR

This paper extends view sampling theory to provide practical guidelines for high-quality view synthesis using local light field fusion, achieving Nyquist-quality results with significantly fewer views and exploring implications for neural radiance fields.

Contribution

It introduces a bound for view sampling density in local light field fusion, enabling high-quality synthesis with far fewer views and discusses sampling strategies for neural radiance fields.

Findings

Achieves Nyquist rate quality with up to 4000x fewer views.

Provides a theoretical bound guiding view sampling density.

Revisits sparse and single-image view synthesis challenges.

Abstract

Capturing and rendering novel views of complex real-world scenes is a long-standing problem in computer graphics and vision, with applications in augmented and virtual reality, immersive experiences and 3D photography. The advent of deep learning has enabled revolutionary advances in this area, classically known as image-based rendering. However, previous approaches require intractably dense view sampling or provide little or no guidance for how users should sample views of a scene to reliably render high-quality novel views. Local light field fusion proposes an algorithm for practical view synthesis from an irregular grid of sampled views that first expands each sampled view into a local light field via a multiplane image scene representation, then renders novel views by blending adjacent local light fields. Crucially, we extend traditional plenoptic sampling theory to derive a bound that specifies precisely how densely users should sample views of a given scene when using our algorithm. We achieve the perceptual quality of Nyquist rate view sampling while using up to 4000x fewer views. Subsequent developments have led to new scene representations for deep learning with view synthesis, notably neural radiance fields, but the problem of sparse view synthesis from a small number of images has only grown in importance. We reprise some of the recent results on sparse and even single image view synthesis, while posing the question of whether prescriptive sampling guidelines are feasible for the new generation of image-based rendering algorithms.