BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Tokyo
X-LIC-LOCATION:Asia/Tokyo
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:JST
DTSTART:18871231T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260817T171532Z
LOCATION:Hall B5 (1)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241206T113100
DTEND;TZID=Asia/Tokyo:20241206T114300
UID:siggraphasia_SIGGRAPH Asia 2024_sess142_papers_700@linklings.com
SUMMARY:Style-NeRF2NeRF: 3D Style Transfer from Style-Aligned Multi-View I
 mages
DESCRIPTION:Haruo Fujiwara (University of Tokyo) and Yusuke Mukuta and Tat
 suya Harada (University of Tokyo, RIKEN AIP)\n\nWe propose a simple yet ef
 fective pipeline for stylizing a 3D scene, harnessing the power of 2D imag
 e diffusion models. Given a NeRF model reconstructed from a set of multi-v
 iew images, we perform 3D style transfer by refining the source NeRF model
  using stylized images generated by a style-aligned image-to-image diffusi
 on model.\nGiven a target style prompt, we first generate perceptually sim
 ilar multi-view images by leveraging a depth-conditioned diffusion model w
 ith an attention-sharing mechanism. Next, based on the stylized multi-view
  images, we propose to guide the style transfer process with the sliced Wa
 sserstein loss based on the feature maps extracted from a pre-trained CNN 
 model.\nOur pipeline consists of decoupled steps, allowing users to test v
 arious prompt ideas and preview the stylized 3D result before proceeding t
 o the NeRF fine-tuning stage.\nWe demonstrate that our method can transfer
  diverse artistic styles to real-world 3D scenes with competitive quality.
 \n\nRegistration Category: Full Access, Full Access Supporter\n\nLanguage 
 Format: English Language\n\nSession Chair: Maria Larsson (The University o
 f Tokyo)\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_700&sess=sess142
END:VEVENT
END:VCALENDAR