BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260805T205608Z
LOCATION:West Hall Lobby
DTSTART;TZID=America/Los_Angeles:20260723T090000
DTEND;TZID=America/Los_Angeles:20260723T173000
UID:siggraph_SIGGRAPH 2026_sess460@linklings.com
SUMMARY:Posters: Animation & Simulation
DESCRIPTION:16. A DEM-Integrated Stream-Function Solver for Simulating San
 d Motion in Liquid–Bubble Flows\n\nWe present a DEM-integrated stream-func
 tion solver for simulating sand motion in liquid-bubble flows. By coupling
  a stream-function fluid solver with DEM, our method models liquid forces 
 and collisions with moving bubble interfaces, producing plausible underwat
 er sand transport and interface-followi...\n\n\nYuma Kurosawa and Yukino M
 asubuchi (Hosei University) and Syuhei Sato (Hosei University, Prometech C
 G Research)\n---------------------\n17. Appearance-Invariant Detection of 
 Suggestive Motion via Laban Movement Descriptors\n\nWe present a privacy-p
 reserving moderation pipeline that flags suggestive motion in online virtu
 al environments from body movement alone -- no pixels, clothing, or body s
 hape -- by classifying interpretable Laban Movement Analysis descriptors c
 omputed from 3D skeletons, matching a fine-tuned deep vi...\n\n\nJaehoon A
 hn, Jeonghan Kong, and Moon-Ryul Jung (Sogang University)\n---------------
 ------\n18. BlendAnything: A Blender Plugin for Cross-Topology Motion Blen
 ding\n\nBlendAnything is a Blender plugin that enables cross-topology moti
 on blending through a skeleton-agnostic diffusion autoencoding backend to 
 interpolate between heterogeneous character structures. The system empower
 s animators to seamlessly transition motions across diverse skeletons, suc
 h as crabs a...\n\n\nLuca Cazzola (University of Trento) and Giulia Martin
 elli and Nicola Conci (University of Trento, CNIT)\n---------------------\
 n19. Deploying Speech-Driven 3D Facial Animation in Unreal Engine for Prod
 uction-Ready Digital Humans\n\nSpeech-driven facial animation has advanced
  rapidly, yet most methods remain incompatible with production pipelines. 
 We bridge this gap by enabling speech-driven animation in Unreal Engine us
 ing ARKit. We build a dataset, retrain state-of-the-art models, and develo
 p a modular plugin for controllable...\n\n\nAlessandro Busacchi, Kazi Inja
 mamul Haque, and Zerrin Yumak (Utrecht University)\n---------------------\
 n20. Fast Modal Dynamics for Hairy and Scaly Surfaces on the GPU\n\nWe pre
 sent a reduced order method for finite element simulation of almost a mill
 ion short furs, spikes or scales attached to a deforming mesh, in real tim
 e (28FPS).\n\n\nYixin Lok (University of Toronto) and David I.W. Levin (Un
 iversity of Toronto, NVIDIA)\n---------------------\n21. From State Recons
 truction to State Interpretation: S2VMO, an Agentic Voxel-Based Digital Tw
 in for Space Habitat Structural Monitoring\n\nMonitoring the structural st
 ate of space habitats is difficult where sensing is sparse and conditions 
 are too complex for routine inspection, demanding an active, context-aware
  expert in the loop. S2VMO is a pilot framework that introduces an agentic
  voxel-based digital twin, which combines finite e...\n\n\nFarshad Askari,
  Shiyun Wen, Jian Tao, Praneet Surabhi, and Wei Yan (Texas A&M University)
 \n---------------------\n22. Geometry-Aware Table Tennis Racket Pose Estim
 ation via Multi-View Consistency and Racket Priors\n\nWe present a geometr
 y-driven framework for estimating table tennis racket pose under occlusion
 . By enforcing cross-view consistency, our method achieves accurate and st
 able pose estimation. An interactive editing tool is further introduced to
  generate high-quality ground truth, improving physical pl...\n\n\nSzu-Han
  Wu, Pei-Hsin Huang, Pu Ching, Hung-Kuo Chu, and Min-Chun Hu (National Tsi
 ng Hua University)\n---------------------\n23. Learning Surfing-like Balan
 ce without Water Simulation\n\nWe present a method for learning surfing-li
 ke balance control without water simulation, enabling a humanoid to mainta
 in stable balance on a moving board under wave conditions despite being tr
 ained entirely in non-fluid environments.\n\n\nHauk Nam, Changho Lee, and 
 Yoonsang Lee (Hanyang University)\n---------------------\n24. Multi Layere
 d RBD Simulation\n\nI present a multi-layered RBD workflow that enables de
 struction simulations to be refined through additional simulation layers w
 hile preserving approved motion. Fracture detail and art-direction changes
  can be incorporated incrementally without rebuilding the original simulat
 ion.\n\n\nJAE JUN YI (Independent, Pixar Animation Studios)\n-------------
 --------\n25. OmniBody: Discretized Local Constraints for Real-Time Destru
 ctible and Flexible Soft Bodies Simulation\n\nOmniBody introduces a scalab
 le GPU framework for real-time soft-body simulation that enables global de
 formation and emergent anisotropic fracture through local interactions of 
 oriented ellipsoidal particles, eliminating the need for global solvers or
  complex re-meshing while ensuring high-fidelity, ...\n\n\nJewoo Shin and 
 Jinah Park (Korea Advanced Institute of Science and Technology (KAIST))\n-
 --------------------\n26. PressureBlush: Real-Time Blanching and Hyperemia
  Shading from Neural Deformation Signals\n\nPressureBlush presents a real-
 time deformation-to-appearance framework that maps self-contact neural def
 ormation signals to shader-computed blanching and hyperemia, producing phy
 siologically motivated skin-color responses without additional runtime sim
 ulation, precomputed textures, or artist-authore...\n\n\nWoonjung Kim (Nex
 on Games) and Haemin Kim and Junyong Noh (Visual Media Lab, KAIST)\n------
 ---------------\n27. SeatSwap: Learning-Assisted Differentiable Kinematics
  for Human-Object Interaction Adaptation in Production-Scale Character Ani
 mation\n\nSeatSwap solves motion capture data scarcity through a learning-
 assisted, differentiable biomechanical optimization framework that automat
 ically adapts human animations to arbitrary 3D furniture geometries, unloc
 king limitless synthetic data generation at production scale.\n\n\nJasmine
  Liang, Stephen Mylabathula, Shai Segev, Priyamvad Ravindra Deshmukh, Ying
  Liu, and Amaury Aubel (Meta; Reality Labs, Meta)\n\nRegistration Category
 : Full Conference Supporter, Full Conference, Experience
END:VEVENT
END:VCALENDAR