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:20260114T163643Z
LOCATION:Meeting Room C4.11\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231213T161500
DTEND;TZID=Australia/Melbourne:20231213T162500
UID:siggraphasia_SIGGRAPH Asia 2023_sess125_papers_769@linklings.com
SUMMARY:Shadow Harmonization for Realistic Compositing
DESCRIPTION:Lucas Valença and Jinsong Zhang (Université Laval), Michaël Gh
 arbi and Yannick Hold-Geoffroy (Adobe), and Jean-François Lalonde (Univers
 ité Laval)\n\nCompositing virtual objects into real background images requ
 ires one to carefully match the scene's camera parameters, surface geometr
 y, textures, and lighting to obtain plausible renderings.\nRecent learning
  approaches have shown many scene properties can be estimated from images,
  resulting in robust automatic single-image compositing systems, but many 
 challenges remain.\nIn particular, interactions between real and synthetic
  shadows are not handled gracefully by existing methods, which typically a
 ssume a shadow-free background. \nAs a result, they tend to generate doubl
 e shadows when the synthetic object's cast shadow overlaps a background sh
 adow, and ignore shadows from the background that should be cast onto the 
 synthetic object. \nIn this paper, we present a compositing method for out
 door scenes that addresses these issues and produces realistic cast shadow
 s.\nThis requires identifying existing shadows, including soft shadow boun
 daries, then reasoning about the ambiguity of unknown ground albedo and sc
 ene lighting to match the color and intensity of shaded areas.\nUsing supe
 rvision from shadow removal and detection datasets, we propose a generativ
 e adversarial pipeline and improved composition equations that simultaneou
 sly handle both shadow interaction scenarios. \nWe evaluate our method on 
 challenging, real outdoor images from multiple distributions and datasets.
 \nQuantitative and qualitative comparisons show our approach produces more
  realistic results than existing alternatives.\n\nRegistration Category: F
 ull Access\n\nSession Chair: Michael Gharbi (Reve AI, Massachusetts Instit
 ute of Technology (MIT))\n\n
URL:https://asia.siggraph.org/2023/full-program?id=papers_769&sess=sess125
END:VEVENT
END:VCALENDAR