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:20260805T205453Z
LOCATION:Room 408 B
DTSTART;TZID=America/Los_Angeles:20260722T104500
DTEND;TZID=America/Los_Angeles:20260722T123500
UID:siggraph_SIGGRAPH 2026_sess153@linklings.com
SUMMARY:Reconstruction
DESCRIPTION:Surface Reconstruction from Unoriented Points via Green’s 3rd 
 Identity\n\nG3R reconstructs surfaces from unoriented point clouds by usin
 g Green's 3rd identity to recover globally consistent normals by solving a
  well-conditioned linear system. It removes the tangential nullspace that 
 weakens prior winding-number-based methods and delivers accurate orientati
 ons and higher-q...\n\n\nZhonghao Wu (University of Science and Technology
  of China, Shanghai Innovation Institute) and Dong Xiao and Renjie Chen (U
 niversity of Science and Technology of China)\n---------------------\nA Ba
 yesian Approach for Task-Specific Next-Best-View Selection with Uncertain 
 Geometry\n\nWe introduce a Bayesian framework for task-specific next-best-
 view selection in 3D reconstruction. Unlike general uncertainty-reduction 
 methods, we selectively reduce uncertainty only in regions critical to the
  downstream task via task-specific acquisition functions on posterior dist
 ributions over i...\n\n\nJingsen Zhu (Cornell University), Silvia Sellán (
 Columbia University), and Alexander Terenin (Cornell University)\n--------
 -------------\nCaRaFe: Camera-Radar Radiance Fields for Scene Reconstructi
 on\n\nCaRaFe is the first method to jointly reconstruct 3D scenes from bot
 h camera and radar data using a multi-modal shared neural geometry represe
 ntation. It can reconstruct large-scale urban environments from a single d
 rive-through and synthesize measurements from both sensors at novel viewpo
 ints, leve...\n\n\nDavid Borts (ETH Zürich); Julian Ost (Princeton Univers
 ity, Torc Robotics); Shamik Basu (INRIA); Tim Broedermann (ETH Zürich); An
 drea Ramazzina (Mercedes-Benz, Technical University of Munich); Christos S
 akaridis (ETH Zürich); and Mario Bijelic and Felix Heide (Princeton Univer
 sity, Torc Robotics)\n---------------------\nPixels2Peaks: Converting Terr
 ain Images to Heightmaps\n\nPixels2Peaks automates the creation of 3D bare
 -earth terrain heightmaps from a single 2D photograph. Our method first es
 timates camera parameters and visible geometry; then uses a guided diffusi
 on model to plausibly synthesize occluded regions. The result maintains ge
 omorphological, hydrological, an...\n\n\nAryamaan Jain (INRIA, Université 
 Côte d'Azur); James Gain (University of Cape Town); and Guillaume Cordonni
 er (INRIA, Université Côte d'Azur)\n---------------------\nMegaNorm: Local
  Patches Embedding for Efficient and Robust Point Normal Orientation at Su
 per-Large Scale\n\nMegaNorm is a fast, accurate, and robust method for ori
 enting normals in non-watertight, scene-scale point clouds, scaling to ~50
 M points in under 10 minutes. It uses two feed-forward networks: one orien
 ts local patches in parallel, while a lightweight MLP evaluates pairwise c
 onsistency. Finally, a ...\n\n\nZhuodong Li, Zengke Liu, Fei Hou, Xuhui Ch
 en, and Wencheng Wang (Institute of Software, Chinese Academy of Sciences;
  University of Chinese Academy of Sciences) and Ying He (Nanyang Technolog
 ical University)\n---------------------\nResolution Where It Counts: Hash-
 based GPU-Accelerated 3D Reconstruction via Variance-Adaptive Voxel Grids\
 n\nWe propose a variance-adaptive voxel grid that adjusts size based on lo
 cal SDF variance. Using a flat spatial hash table, it achieves constant-ti
 me access, memory efficiency, and full GPU parallelism. Additionally, a pa
 rallel quad-tree structure supports GPU-accelerated Gaussian Splatting ren
 dering,...\n\n\nLorenzo De Rebotti, Emanuele Giacomini, Giorgio Grisetti, 
 and Luca Di Giammarino (University of Rome La Sapienza) and Lorenzo De Reb
 otti\n---------------------\nArtiFixer: Enhancing and Extending 3D Reconst
 ruction with Auto-Regressive Diffusion Models\n\nArtiFixer enhances and ex
 tends 3D reconstructions efficiently and scalably. Given an initial recons
 truction and reference views, it autoregressive generates novel content co
 nsistent with existing observations. ArtiFixer can produce hundreds of nov
 el views in a single inference pass or serve as pseud...\n\n\nRiccardo De 
 Lutio (NVIDIA); Tobias Fischer (NVIDIA, ETH Zürich); Yen-Yu Chang (NVIDIA,
  Cornell University); Yuxuan Zhang and Zhangjie Wu (NVIDIA); Xuanchi Ren a
 nd Tianchang Shen (NVIDIA, University of Toronto); and Katarína Tóthová, Z
 an Gojcic, and Haithem Turki (NVIDIA)\n---------------------\nReconstructi
 on - Interactive Discussion\n\nAfter the summary presentations, attendees 
 will participate in an interactive discussion. Outside the room will be a 
 series of poster boards for authors to gather around with the audience. Au
 thors are invited to bring any material related to their paper that could 
 instigate further conversation such...\n\n\nInterest Area: Research & Educ
 ation\n\nKeyword: Artificial Intelligence/Machine Learning, Generative AI,
  Geometry, Modeling, Rendering, Simulation\n\nRegistration Category: Full 
 Conference Supporter, Full Conference\n\nSession Chair: Oded Stein (Techni
 on, University of Southern California)
END:VEVENT
END:VCALENDAR