BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240214T070245Z
LOCATION:Meeting Room C4.11\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231213T145000
DTEND;TZID=Australia/Melbourne:20231213T150500
UID:siggraphasia_SIGGRAPH Asia 2023_sess128_papers_227@linklings.com
SUMMARY:CamP: Camera Preconditioning for Neural Radiance Fields
DESCRIPTION:Technical Papers\n\nKeunhong Park, Phillip Henzler, Ben Milden
 hall, Jonathan T. Barron, and Ricardo Martin-Brualla (Google Research)\n\n
 Neural Radiance Fields (NeRF) can be optimized to obtain high-fidelity 3D 
 scene reconstructions of objects and large-scale scenes. However, NeRFs re
 quire accurate camera parameters as input --- inaccurate camera parameters
  result in blurry renderings. Extrinsic and intrinsic camera parameters ar
 e usually estimated using Structure-from-Motion (SfM) methods as a pre-pro
 cessing step to NeRF, but these techniques rarely yield perfect estimates.
  Thus, prior works have proposed jointly optimizing camera parameters alon
 gside a NeRF, but these methods are prone to local minima in challenging s
 ettings.\n\nIn this work, we analyze how different camera parameterization
 s affect this joint optimization problem, and observe that standard parame
 terizations exhibit large differences in magnitude with respect to small p
 erturbations, which can lead to an ill-conditioned optimization problem. W
 e propose using a proxy problem to compute a whitening transform that elim
 inates the correlation between camera parameters and normalizes their effe
 cts, and we propose to use this transform as a preconditioner for the came
 ra parameters during joint optimization.\n\nOur preconditioned camera opti
 mization significantly improves reconstruction quality on scenes from the 
 Mip-NeRF 360 dataset: we reduce error rates (RMSE) by 67% compared to stat
 e-of-the-art NeRF approaches that do not optimize for cameras like Zip-NeR
 F, and by 29% relative to state-of-the-art joint optimization approaches u
 sing the camera parameterization of SCNeRF. Our approach is easy to implem
 ent, does not significantly increase runtime, can be applied to a wide var
 iety of camera parameterizations, and can straightforwardly be incorporate
 d into other NeRF-like models.\n\nRegistration Category: Full Access\n\nSe
 ssion Chair: Jianfei Cai (Monash University)
URL:https://asia.siggraph.org/2023/full-program?id=papers_227&sess=sess128
END:VEVENT
END:VCALENDAR