Overview
Syllabus
Outline
What is Novel View Synthesis?
Stereo Magnification: Learning view synthesis using multiplane images, Zhou et al.
Stereo Magnification: Multiplane Image
DeepMVS: Learning multi-view stereopsis, Huang et al.
DeepMVS: Patch Matching
DeepMVS: Inter-Volume Feature Aggregation
DeepMVS: Disparity Maps
What is trying to be solved?
Proposed Solution
Proposed Architecture
Estimating the Depth Probability Volume
Depth Probability Distributions and Discontinuities
Estimating Depth Probability of the Novel Vie
Synthesis of a Novel View
Artifacts Present After Synthesis
Patch Warping
Refinement Network Training
Refinement Example
Evaluation
Quantitative Results
Comparison to Stereo Magnification
Comparison to Soft3D
Limitations
Conclusion
Taught by
UCF CRCV