README.md

Texture and Structure in NeRF

This dataset contains randomly placed shapes with procedural textures for the purpose of measuring the correlation of texture frequency and solid structure in NeRF reconstructions.

The dataset is available at gs://kubric-public/data/texture_structure_nerf

Running the worker

sudo docker run --rm --interactive \
  --user $(id -u):$(id -g)         \
  --volume "$(pwd):/kubric"        \
  kubricdockerhub/kubruntu         \
  /usr/bin/python3                 \
  examples/nerf_texture.py

Parameters:

• num_objects How many objects to generate.
• num_frequency_bands How many discrete frequency bands to use.
• min_log_frequency Minimum frequency value in log-scale (base 10).
• max_log_frequency Maximum frequency value in log-scale (base 10).
• num_train_frames How many frames to render in the training split.
• num_validation_frames How many frames to render in the validation split.
• num_test_frames How many frames to render in the testing split.

Output Format

The script directly generates output that can be used as input by JAXNeRF with the 'blender' configuration. The resulting folder structure is:

• [train|val|test]/*.png RGB color images.
• [train|val|test]/*_segmentation.png Segmentation maps indicating which frequency band a pixel belongs to.
• transforms_[train|val|test].json Camera information for each data split.