The data are in the same format as the data used in Nerfies, except for the additional masks. Left cameras are used for training and right cameras are used for testing.
dataset
├── camera
│ └── ${item_id}.json
├── camera-paths
├── rgb
│ ├── ${scale}x
│ └── └── ${item_id}.png
├── mask
│ ├── ${scale}x
│ └── └── ${item_id}.png
├── metadata.json
├── points.npy
├── dataset.json
└── scene.json