notesum.ai

Published at November 18

SpatialDreamer: Self-supervised Stereo Video Synthesis from Monocular Input

cs.CV

cs.AI

Released Date: November 18, 2024

Authors: Zhen Lv¹, Yangqi Long¹, Congzhentao Huang¹, Cao Li¹, Chengfei Lv¹, Hao Ren¹, Dian Zheng¹

Aff.: ¹Alibaba Group

Arxiv: http://arxiv.org/abs/2411.11934v1

Refer to caption

	SSIM $\uparrow$		PSNR $\uparrow$		LPIPS $\downarrow$
	$t$ =5	$t$ =10	$t$ =5	$t$ =10	$t$ =5	$t$ =10
3D-photography [54] $\dagger$	0.641	0.446	11.08	9.533	0.116	0.386
SynSin [67] $\dagger$	0.740	0.645	21.22	19.55	0.087	0.104
AdaMPI [21] $\dagger$	0.869	0.736	27.70	22.65	0.041	0.087
SinMPI [45] $\dagger$	0.898	0.775	28.79	23.21	0.040	0.080
Wang et al. [64]	0.840	0.720	25.35	21.36	0.049	0.095
P-NVS [73] $\dagger$	0.724	0.667	21.98	19.46	0.241	0.279
NVSVDE-Net [4]	0.840		24.31		0.233
CoPoNeRF [24] $\dagger$	0.658	0.632	22.63	21.33	0.171	0.209
MVSplat [14] $\dagger$	0.863		25.89		0.132
Proposed	0.916	0.857	32.26	24.86	0.038	0.049