BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Hong_Kong
X-LIC-LOCATION:Asia/Hong_Kong
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:HKT
DTSTART:19911015T033000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20251218T030653Z
LOCATION:Meeting Room S423+S424\, Level 4
DTSTART;TZID=Asia/Hong_Kong:20251217T145000
DTEND;TZID=Asia/Hong_Kong:20251217T150000
UID:siggraphasia_SIGGRAPH Asia 2025_sess141_papers_1116@linklings.com
SUMMARY:From Rigging to Waving: 3D-Guided Diffusion for Natural Animation 
 of Hand-Drawn Characters
DESCRIPTION:Jie ZHOU, Linzi QU, and Miu-Ling LAM (City University of Hong 
 Kong) and Hongbo FU (Hong Kong University of Science and Technology)\n\nHa
 nd-drawn character animation is a vibrant research area in computer graphi
 cs, presenting unique challenges in achieving geometric consistency while 
 conveying expressive motion details. Traditional skeletal animation method
 s maintain geometric consistency but often struggle with complex non-rigid
  elements like flowing hair and skirts, resulting in unnatural deformation
  and missing secondary dynamics. In contrast, video diffusion models effec
 tively synthesize physics-aware dynamics but suffer from stylized artifact
 s and geometric distortions when applied to stylized drawings due to domai
 n gaps. In this work, we propose a novel hybrid animation system that inte
 grates the strengths of skeletal animation and video diffusion priors.  Th
 e core idea is to generate coarse images from characters retargeted with s
 keletal animations for geometric consistency guidance and to further enhan
 ce these images with video diffusion models in terms of texture details an
 d secondary dynamics. We reformulate the enhancement of coarse images as a
 n inpainting task and propose a domain-adapted diffusion model to refine r
 egions requiring improvement, particularly those involving secondary dynam
 ics, guided by user-provided masks. To further enhance motion realism, we 
 propose a Secondary Dynamics Enhancement strategy during the denoising pro
 cess that incorporates latent features from a pre-trained diffusion model 
 enriched with human motion priors. Additionally, to address unnatural defo
 rmation resulting from hair sticking in skeletal animation, we introduce a
  hair layering modeling method that employs segmentation maps to separate 
 hair from the body in the implicit fields, allowing our system to animate 
 challenging hair-sticking characters more naturally. Through extensive exp
 eriments and a perceptual study, we demonstrate that our system generates 
 high-fidelity animations with realistic dynamics and artistic integrity. T
 he code and more animation results are available in the supplementary mate
 rials.\n\nRegistration Category: Full Access, Full Access Supporter\n\nSes
 sion Chair: Wanchao Su (Monash University)\n\n
END:VEVENT
END:VCALENDAR