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:20260817T171538Z
LOCATION:G510\, G Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241205T130000
DTEND;TZID=Asia/Tokyo:20241205T135000
UID:siggraphasia_SIGGRAPH Asia 2024_sess288@linklings.com
SUMMARY:Stylization & Evaluation
DESCRIPTION:The Technical Communications program at SIGGRAPH Asia serves a
 s an invaluable platform for presenting cutting-edge work that may not nea
 tly align with the Technical Papers session. Attendees can expect to explo
 re fresh and thought-provoking ideas, glean practical insights from real-w
 orld production work, and discover innovative applications spanning variou
 s disciplines, from graphics and vision to AI and VR.\n\nDuring these sess
 ions, leading experts from academia and industry will present their latest
  findings, offering a glimpse into cutting-edge research and development. 
 From geometry and animation to virtual reality and machine learning, atten
 dees can expect to explore a diverse array of topics at the intersection o
 f graphics and other fields.\n\nUnder the overarching theme of Curious Min
 ds, attendees can delve into discussions surrounding innovation, interdisc
 iplinary discovery, and the role of education in shaping the future of tec
 hnology. Whether you’re a seasoned researcher or a curious enthusiast, the
  Technical Communications program promises to offer insights that spark cu
 riosity and inspire new perspectives.\n\nStyleGaussian: Instant 3D Style T
 ransfer with Gaussian Splatting\n\nStyleGaussian is a novel 3D style trans
 fer pipeline that enables instant style transfer while preserving real-tim
 e rendering and strict multi-view consistency.\n\n\nKunhao Liu (Nanyang Te
 chnological University (NTU)); Fangneng Zhan (Max Planck Institute for Inf
 ormatics); Muyu Xu (Nanyang Technological University (NTU)); Christian The
 obalt (Max Planck Institute for Informatics); Ling Shao (UCAS-Terminus AI 
 Lab, UCAS); and Shijian Lu (Nanyang Technological University (NTU))\n-----
 ----------------\nA Practical Style Transfer Pipeline for 3D Animation: In
 sights from Production R&D\n\nWe developed a practical style transfer pipe
 line for creating stylized 3D animation, suitable for real-world productio
 n. This paper presents insights from our R&D process, balancing quality, a
 rtist control, and workload.\n\n\nHideki Todo (Takushoku University); Yuki
  Koyama, Kunihiro Sakai, and Akihiro Komiya (Graphinica, Inc.); and Jun Ka
 to (Arch Inc.)\n---------------------\nAnimateLCM: Computation-Efficient P
 ersonalized Style Video Generation without Personalized Video Data\n\nComp
 utation-efficient personalized style video generation without personalized
  video data, reducing generation time of similarly sized video diffusion m
 odels from 25 seconds to around 1 second while maintaining comparable perf
 ormance.\n\n\nFu-Yun Wang (Chinese University of Hong Kong); Zhaoyang Huan
 g (Avolution AI); Weikang Bian, Xiaoyu Shi, and Keqiang Sun (Chinese Unive
 rsity of Hong Kong); Guanglu Song (SenseTime); Yu Liu (Shanghai Artificial
  Intelligence Laboratory); and Hongsheng Li (Chinese University of Hong Ko
 ng)\n---------------------\nAn Empirical Analysis of GPT-4V’s Performance 
 on Fashion Aesthetic Evaluation\n\nWe examine GPT-4V's zero-shot performan
 ce in evaluating outfit suitability in images, showing good alignment with
  human judgments but difficulties in ranking similarly colored outfits.\n\
 n\nYuki Hirakawa, Takashi Wada, Kazuya Morishita, Ryotaro Shimizu, Takuya 
 Furusawa, Sai Htaung Kham, and Yuki Saito (ZOZO)\n\nRegistration Category:
  Full Access, Full Access Supporter\n\nLanguage Format: English Language\n
 \nSession Chair: Krishna Mullia (Canva Research)
END:VEVENT
END:VCALENDAR