Learning to Synthesize a 4D RGBD Light Field from a Single Image
This work addresses the challenge of light field synthesis for applications in computer vision and graphics, representing an incremental advancement over existing view synthesis methods.
The paper tackles the problem of generating a 4D RGBD light field from a single 2D RGB image, achieving this by using a convolutional neural network pipeline that estimates scene geometry and predicts occluded rays and non-Lambertian effects, with results validated on a new large dataset of over 3300 light fields.
We present a machine learning algorithm that takes as input a 2D RGB image and synthesizes a 4D RGBD light field (color and depth of the scene in each ray direction). For training, we introduce the largest public light field dataset, consisting of over 3300 plenoptic camera light fields of scenes containing flowers and plants. Our synthesis pipeline consists of a convolutional neural network (CNN) that estimates scene geometry, a stage that renders a Lambertian light field using that geometry, and a second CNN that predicts occluded rays and non-Lambertian effects. Our algorithm builds on recent view synthesis methods, but is unique in predicting RGBD for each light field ray and improving unsupervised single image depth estimation by enforcing consistency of ray depths that should intersect the same scene point. Please see our supplementary video at https://youtu.be/yLCvWoQLnms