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:20251218T030656Z
LOCATION:Meeting Room S423+S424\, Level 4
DTSTART;TZID=Asia/Hong_Kong:20251217T110100
DTEND;TZID=Asia/Hong_Kong:20251217T111200
UID:siggraphasia_SIGGRAPH Asia 2025_sess135_papers_1445@linklings.com
SUMMARY:Generating 360° Video is What You Need For a 3D Scene
DESCRIPTION:Zhaoyang Zhang (Yale University, Adobe Research); Yannick Hold
 -Geoffroy and Miloš Hašan (Adobe Research); Ziwen Chen (Oregon State Unive
 rsity); Fujun Luan (Adobe Research); Julie Dorsey (Yale University); and Y
 iwei Hu (Adobe Research)\n\nGenerating 3D scenes is still a challenging ta
 sk due to the lack of readily available scene data. Most existing methods 
 only produce partial scenes and provide limited navigational freedom. We i
 ntroduce a practical and scalable solution that uses 360° video as an inte
 rmediate scene representation, capturing the full-scene context and ensuri
 ng consistent visual content throughout the generation. We propose WorldPr
 ompter, a generative pipeline that synthesizes traversable 3D scenes from 
 text prompts. WorldPrompter incorporates a conditional 360° panoramic vide
 o generator, capable of producing a 128-frame video that simulates a perso
 n walking through and capturing a virtual environment. The resulting video
  is then reconstructed as Gaussian splats by a fast feedforward 3D reconst
 ructor, enabling a true walkable experience within the 3D scene. Experimen
 ts demonstrate that our panoramic video generation model, trained with a m
 ix of image and video data, achieves convincing spatial and temporal consi
 stency for static scenes. This is validated by an average COLMAP matching 
 rate of 94.6%, allowing for high-quality panoramic Gaussian splat reconstr
 uction and improved navigation throughout the scene. Qualitative and quant
 itative results also show it outperforms the state-of-the-art 360° video g
 enerators and 3D scene generation models.\n\nRegistration Category: Full A
 ccess, Full Access Supporter\n\nSession Chair: Or Patashnik (Tel Aviv Univ
 ersity, Snap Research)\n\n
END:VEVENT
END:VCALENDAR