BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Seoul
X-LIC-LOCATION:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:18871231T000000
DTSTART:19881009T020000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230103T035311Z
LOCATION:Room 325-AB\, Level 3\, West Wing
DTSTART;TZID=Asia/Seoul:20221208T140000
DTEND;TZID=Asia/Seoul:20221208T153000
UID:siggraphasia_SIGGRAPH Asia 2022_sess169_papers_171@linklings.com
SUMMARY:Learning to Relight Portrait Images via a Virtual Light Stage and 
 Synthetic-to-Real Adaptation
DESCRIPTION:Technical Communications, Technical Papers\n\nLearning to Reli
 ght Portrait Images via a Virtual Light Stage and Synthetic-to-Real Adapta
 tion\n\nYeh, Nagano, Khamis, Kautz, Liu...\n\nGiven a portrait image of a 
 person and an environment map of the target lighting, portrait relighting 
 aims to re-illuminate the person in the image as if the person appeared in
  an environment with the target lighting. To achieve high-quality results,
  recent methods rely on deep learning. An effective approach is to supervi
 se the training of deep neural networks with a high-fidelity dataset of de
 sired input--output pairs, captured with a light stage. However, acquiring
  such data requires an expensive special capture rig and time-consuming ef
 forts, limiting access to only a few resourceful laboratories. To address 
 the limitation, we propose a new approach that can perform on par with the
  state-of-the-art (SOTA) relighting methods without requiring a light stag
 e. Our approach is based on the realization that a successful relighting o
 f a portrait image depends on two conditions. First, the physics of light 
 transport has to be correct. Second, the output has to be photorealistic. 
 To meet the first condition, we propose to train the relighting network wi
 th training data generated by a virtual light stage that performs physical
 ly-based rendering on various 3D synthetic humans under different environm
 ent maps. To meet the second condition, we develop a novel synthetic-to-re
 al approach to bring photorealism to the relighting network output. In add
 ition to achieving SOTA results, our approach offers several advantages ov
 er the prior methods, including controllable glares on glasses and more te
 mporally-consistent results for relighting videos.\n\nRegistration Categor
 y: FULL ACCESS, ON-DEMAND ACCESS\n\nLanguage: ENGLISH\n\nFormat: IN-PERSON
 , ON-DEMAND
URL:https://sa2022.siggraph.org/en/full-program/?id=papers_171&sess=sess16
 9
END:VEVENT
END:VCALENDAR