BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CAIL//Events//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:CAIL Events
BEGIN:VEVENT
UID:2025-06-04-lior-yariv@cail.columbia.edu
DTSTAMP:20260823T060943Z
DTSTART:20250604T170000Z
DTEND:20250604T180000Z
SUMMARY:Vision Seminar: Lior Yariv - Learning Signed Distance Representati
 ons for 3D Modeling
LOCATION:CEPSR 620
DESCRIPTION:A central question in learning 3D scenes is choosing an approp
 riate and efficient scene representation\, one that faithfully represents 
 the scene’s underlying geometry. The goal of this talk is to present Sig
 ned Distance Functions (SDFs) as a core representation for 3D modeling. SD
 Fs offer a flexible and expressive way to model a wide variety of closed s
 urfaces with complex topology\, while also allowing for straightforward su
 rface extraction. Moreover\, the geometric information encoded in SDFs fac
 ilitates accurate rendering and supports favorable reconstruction. We pres
 ent the potential of the SDF representation in addressing key problems in 
 computer vision and computer graphics\, focusing on two fundamental tasks:
  multi-view surface reconstruction and 3D generation. For multi-view surfa
 ce reconstruction\, we introduce novel methods for learning SDF-based repr
 esentations directly from 2D images\, by integrating SDFs into differentia
 ble volume rendering pipelines. Towards 3D generation\, we present a scala
 ble and expressive surface representation tailored for training flow-based
  generative models on large-scale 3D shape datasets. Together\, these cont
 ributions demonstrate the versatility and effectiveness of SDFs in address
 ing key challenges in learning meaningful 3D representations under differe
 nt supervision settings.\n\nhttps://cail.columbia.edu/events/2025-06-04-li
 or-yariv
URL:https://cail.columbia.edu/events/2025-06-04-lior-yariv
END:VEVENT
END:VCALENDAR