BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240214T070240Z
LOCATION:Darling Harbour Theatre\, Level 2 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231212T093000
DTEND;TZID=Australia/Melbourne:20231212T124500
UID:siggraphasia_SIGGRAPH Asia 2023_sess209_papers_812@linklings.com
SUMMARY:Diffusion-based Holistic Texture Rectification and Synthesis
DESCRIPTION:Technical Papers\n\nGuoqing Hao (University of Tsukuba, Nation
 al Institute of Advanced Industrial Science and Technology); Satoshi Iizuk
 a (University of Tsukuba); Kensho Hara (National Institute of Advanced Ind
 ustrial Science); Edgar Simo-Serra (Waseda University); Hirokatsu Kataoka 
 (National Institute of Advanced Industrial Science); and Kazuhiro Fukui (U
 niversity of Tsukuba)\n\nWe present a novel framework for rectifying occlu
 sions and distortions in degraded texture samples from natural images. Tra
 ditional texture synthesis approaches focus on generating textures from pr
 istine samples, which necessitate meticulous preparation by humans and are
  often unattainable in most natural images. These challenges stem from the
  frequent occlusions and distortions of texture samples in natural images 
 due to obstructions and variations in object surface geometry. To address 
 these issues, we propose a framework that synthesizes holistic textures fr
 om degraded samples in natural images, extending the applicability of exem
 plar-based texture synthesis techniques. Our framework utilizes a conditio
 nal Latent Diffusion Model (LDM) with a novel occlusion-aware latent trans
 former. This latent transformer not only effectively encodes texture featu
 res from partially-observed samples necessary for the generation process o
 f the LDM, but also explicitly captures long-range dependencies in samples
  with large occlusions. To train our model, we introduce a method for gene
 rating synthetic data by applying geometric transformations and free-form 
 mask generation to clean textures. Experimental results demonstrate that o
 ur framework significantly outperforms existing methods both quantitativel
 y and quantitatively. Furthermore, we conduct comprehensive ablation studi
 es to validate the different components of our proposed framework. Results
  are corroborated by a perceptual user study which highlights the efficien
 cy of our proposed approach.\n\nRegistration Category: Full Access, Enhanc
 ed Access, Trade Exhibitor, Experience Hall Exhibitor
URL:https://asia.siggraph.org/2023/full-program?id=papers_812&sess=sess209
END:VEVENT
END:VCALENDAR