English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This paper introduces a novel paradigm for the generalizable neural radiancefield (NeRF). Previous generic NeRF methods combine multiview stereo techniqueswith image-based neural rendering for generalization, yielding impressiveresults, while suffering from three issues. First, occlusions often result ininconsistent feature matching. Then, they deliver distortions and artifacts ingeometric discontinuities and locally sharp shapes due to their individualprocess of sampled points and rough feature aggregation. Third, theirimage-based representations experience severe degradations when source viewsare not near enough to the target view. To address challenges, we propose thefirst paradigm that constructs the generalizable neural field based onpoint-based rather than image-based rendering, which we call the Generalizableneural Point Field (GPF). Our approach explicitly models visibilities bygeometric priors and augments them with neural features. We propose a novelnonuniform log sampling strategy to improve both rendering speed andreconstruction quality. Moreover, we present a learnable kernel spatiallyaugmented with features for feature aggregations, mitigating distortions atplaces with drastically varying geometries. Besides, our representation can beeasily manipulated. Experiments show that our model can deliver bettergeometries, view consistencies, and rendering quality than all counterparts andbenchmarks on three datasets in both generalization and finetuning settings,preliminarily proving the potential of the new paradigm for generalizable NeRF.

Learning Robust Generalizable Radiance Field with Visibility and Feature Augmented Point Representation