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

3D scene stylization aims at generating stylized images of the scene fromarbitrary novel views following a given set of style examples, while ensuringconsistency when rendered from different views. Directly applying methods forimage or video stylization to 3D scenes cannot achieve such consistency. Thanksto recently proposed neural radiance fields (NeRF), we are able to represent a3D scene in a consistent way. Consistent 3D scene stylization can beeffectively achieved by stylizing the corresponding NeRF. However, there is asignificant domain gap between style examples which are 2D images and NeRFwhich is an implicit volumetric representation. To address this problem, wepropose a novel mutual learning framework for 3D scene stylization thatcombines a 2D image stylization network and NeRF to fuse the stylizationability of 2D stylization network with the 3D consistency of NeRF. We firstpre-train a standard NeRF of the 3D scene to be stylized and replace its colorprediction module with a style network to obtain a stylized NeRF. It isfollowed by distilling the prior knowledge of spatial consistency from NeRF tothe 2D stylization network through an introduced consistency loss. We alsointroduce a mimic loss to supervise the mutual learning of the NeRF stylemodule and fine-tune the 2D stylization decoder. In order to further make ourmodel handle ambiguities of 2D stylization results, we introduce learnablelatent codes that obey the probability distributions conditioned on the style.They are attached to training samples as conditional inputs to better learn thestyle module in our novel stylized NeRF. Experimental results demonstrate thatour method is superior to existing approaches in both visual quality andlong-range consistency.

StylizedNeRF: Consistent 3D Scene Stylization as Stylized NeRF via 2D-3D Mutual Learning