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

Recent methods for neural surface representation and rendering, for exampleNeuS, have demonstrated the remarkably high-quality reconstruction of staticscenes. However, the training of NeuS takes an extremely long time (8 hours),which makes it almost impossible to apply them to dynamic scenes with thousandsof frames. We propose a fast neural surface reconstruction approach, calledNeuS2, which achieves two orders of magnitude improvement in terms ofacceleration without compromising reconstruction quality. To accelerate thetraining process, we parameterize a neural surface representation bymulti-resolution hash encodings and present a novel lightweight calculation ofsecond-order derivatives tailored to our networks to leverage CUDA parallelism,achieving a factor two speed up. To further stabilize and expedite training, aprogressive learning strategy is proposed to optimize multi-resolution hashencodings from coarse to fine. We extend our method for fast training ofdynamic scenes, with a proposed incremental training strategy and a novelglobal transformation prediction component, which allow our method to handlechallenging long sequences with large movements and deformations. Ourexperiments on various datasets demonstrate that NeuS2 significantlyoutperforms the state-of-the-arts in both surface reconstruction accuracy andtraining speed for both static and dynamic scenes. The code is available at ourwebsite: https://vcai.mpi-inf.mpg.de/projects/NeuS2/ .

NeuS2: Fast Learning of Neural Implicit Surfaces for Multi-view Reconstruction