No description
Find a file
2020-10-15 22:12:48 -04:00
configs clean code 2020-10-15 10:19:58 -04:00
demo update demo 2020-10-15 18:44:08 -04:00
.gitignore clean code 2020-10-12 11:05:53 -04:00
data_loader_split.py clean code 2020-10-12 11:05:53 -04:00
ddp_model.py clean code 2020-10-12 11:05:53 -04:00
ddp_test_nerf.py clean code 2020-10-12 11:05:53 -04:00
ddp_train_nerf.py clean code 2020-10-15 22:12:48 -04:00
environment.yml add environment 2020-10-11 22:33:12 -04:00
nerf_network.py clean code 2020-10-15 10:19:58 -04:00
nerf_sample_ray_split.py clean code 2020-10-15 19:29:30 -04:00
README.md Update README.md 2020-10-15 21:56:30 -04:00
utils.py clean code 2020-10-15 19:29:30 -04:00

NeRF++

Codebase for arXiv preprint "NeRF++: Analyzing and Improving Neural Radiance Fields"

  • Work with 360 capture of large-scale unbounded scenes.
  • Support multi-gpu training and inference with PyTorch DistributedDataParallel (DDP).
  • Optimize per-image autoexposure (experimental feature).

Demo

Data

  • Download our preprocessed data from tanks_and_temples, lf_data.
  • Put the data in the sub-folder data/ of this code directory.
  • Data format.
    • Each scene consists of 3 splits: train/test/validation.
    • Intrinsics and poses are stored as flattened 4x4 matrices (row-major).
    • Pixel coordinate of an image's upper-left corner is (column, row)=(0, 0).
    • Poses are camera-to-world, not world-to-camera transformations.
    • Opencv camera coordinate system is adopted, i.e., x--->right, y--->down, z--->scene.
    • To convert camera poses between Opencv and Opengl conventions, the following code snippet can be used for both Opengl2Opencv and Opencv2Opengl.
      import numpy as np
      def convert_pose(C2W):
          flip_yz = np.eye(4)
          flip_yz[1, 1] = -1
          flip_yz[2, 2] = -1
          C2W = np.matmul(C2W, flip_yz)
          return C2W
      
    • Scene normalization: move the average camera center to origin, and put all the camera centers inside the unit sphere.

Create environment

conda env create --file environment.yml
conda activate nerf-ddp

Training (Use all available GPUs by default)

python ddp_train_nerf.py --config configs/tanks_and_temples/tat_training_truck.txt

Testing (Use all available GPUs by default)

python ddp_test_nerf.py --config configs/tanks_and_temples/tat_training_truck.txt \
                        --render_splits test,camera_path

Note: due to restriction imposed by torch.gather function, please make sure the number of pixels in each image is divisible by the number of GPUs if you render images parallelly.

Citation

Plese cite our work if you use the code.

@article{kaizhang2020,
    author    = {Kai Zhang and Gernot Riegler and Noah Snavely and Vladlen Koltun},
    title     = {NeRF++: Analyzing and Improving Neural Radiance Fields},
    journal   = {arXiv:2010.07492},
    year      = {2020},
}