• Stars
    star
    123
  • Rank 283,235 (Top 6 %)
  • Language
    Python
  • License
    MIT License
  • Created about 2 years ago
  • Updated over 1 year ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Official code release for VoxGRAF: Fast 3D-Aware Image Synthesis with Sparse Voxel Grids

VoxGRAF


This repository contains official code for the paper VoxGRAF: Fast 3D-Aware Image Synthesis with Sparse Voxel Grids.

You can find detailed usage instructions for training your own models and using pre-trained models below.

If you find our code or paper useful, please consider citing

@inproceedings{Schwarz2022NEURIPS,
  title = {VoxGRAF: Fast 3D-Aware Image Synthesis with Sparse Voxel Grids},
  author = {Schwarz, Katja and Sauer, Axel and Niemeyer, Michael and Liao, Yiyi and Geiger, Andreas},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2022}
}

Installation

First you have to make sure that you have all dependencies in place. The simplest way to do so, is to use anaconda.

You can create and activate an anaconda environment called voxgraf using

conda env create -f environment.yml
conda activate voxgraf

CUDA extension installation

Install pre-compiled CUDA extensions by running

./scripts/build_wheels.sh

Or install them individually by running

pip install dist/stylegan3_cuda-0.0.0-cp39-cp39-linux_x86_64.whl
pip install dist/svox2-voxgraf-0.0.1.dev0+sphtexcub.lincolor.fast-cp39-cp39-linux_x86_64.whl
pip install dist/MinkowskiEngine-0.5.4-cp39-cp39-linux_x86_64.whl       # optional, only required when training with minkowski sparse convolutions

In case the wheels do not work for you, you can also install the extensions from source. For this please check the original repos: Stylegan-3, Minkowski Engine and for our version of Plenoxels follow the instructions here.

Pretrained models

To download the pretrained models run

./scripts/download_pretrained_models.sh

Evaluate pretrained models

# generate a video with 1x2 samples and interpolations between 2 keyframes each
python gen_video.py --network pretrained_models/ffhq256.pkl --seeds 0-3 --grid 1x2 --num-keyframes 2 --output ffhq_256_samples/video.mp4 --trunc=0.5

# generate grids of 3x4 samples and their depths
python gen_images.py --network pretrained_models/ffhq256.pkl --seeds 0-23 --grid 3x4 --outdir ffhq_256_samples --save_depth true --trunc=0.5

Train custom models

Download the data

Download FFHQ, AFHQ and Carla.

Preparing the data

To prepare the data at the required resolutions you can run

./scripts/make_dataset.sh /PATH/TO/IMAGES data/{DATASET_NAME}.json data/{DATASET_NAME} 32,64,128,256

This will create the datasets in data/{DATASET_NAME}_{RES}.zip.

Train models progressively

# Train a model on FFHQ progressively starting at image resolution 32x32 with voxel grid resolution 32x32x32
python train.py --outdir training-runs --gpus 8 --data data/ffhq_32.zip --batch 64 --grid-res 32
python train.py --outdir training-runs --gpus 8  --data data/ffhq_64.zip --batch 64 --grid-res 32 --resume /PATH/TO/32-IMG-32-GRID-MODEL                                    # Next stage
python train.py --outdir training-runs --gpus 8  --data data/ffhq_64.zip --batch 64 --grid-res 64 --resume /PATH/TO/64-IMG-32-GRID-MODEL                                    # Next stage
python train.py --outdir training-runs --gpus 8  --data data/ffhq_128.zip --batch 64 --grid-res 64 --lambda_vardepth 1e-3 --resume /PATH/TO/64-IMG-64-GRID-MODEL            # Next stage
python train.py --outdir training-runs --gpus 8  --data data/ffhq_128.zip --batch 32 --grid-res 128 --lambda_vardepth 1e-3 --resume /PATH/TO/128-IMG-64-GRID-MODEL          # Next stage
python train.py --outdir training-runs --gpus 8  --data data/ffhq_256.zip --batch 32 --grid-res 128 --lambda_vardepth 1e-3 --resume /PATH/TO/128-IMG-128-GRID-MODEL         # Next stage

# Train a model on Carla at image resolution 32x32 with voxel grid resolution 32x32x32
python train.py --outdir training-runs --gpus 8  --data data/ffhq_32.zip --batch 64 --grid-res 32 --n-refinement 0 --use_bg False --lambda_sparsity 1e-8
python train.py --outdir training-runs --gpus 8  --data data/ffhq_64.zip --batch 64 --grid-res 32 --n-refinement 0 --use_bg False --lambda_sparsity 1e-8 --resume /PATH/TO/32-IMG-32-GRID-MODEL                                    # Next stage
python train.py --outdir training-runs --gpus 8  --data data/ffhq_64.zip --batch 64 --grid-res 64 --n-refinement 0 --use_bg False --lambda_sparsity 1e-8 --resume /PATH/TO/64-IMG-32-GRID-MODEL                                    # Next stage
python train.py --outdir training-runs --gpus 8  --data data/ffhq_128.zip --batch 64 --grid-res 64 --n-refinement 0 --use_bg False --lambda_sparsity 1e-8 --lambda_vardepth 1e-3 --resume /PATH/TO/64-IMG-64-GRID-MODEL            # Next stage
python train.py --outdir training-runs --gpus 8  --data data/ffhq_128.zip --batch 32 --grid-res 128 --n-refinement 0 --use_bg False --lambda_sparsity 1e-8 --lambda_vardepth 1e-3 --resume /PATH/TO/128-IMG-64-GRID-MODEL          # Next stage

More Repositories

1

sdfstudio

A Unified Framework for Surface Reconstruction
Python
1,861
star
2

occupancy_networks

This repository contains the code for the paper "Occupancy Networks - Learning 3D Reconstruction in Function Space"
Python
1,454
star
3

giraffe

This repository contains the code for the CVPR 2021 paper "GIRAFFE: Representing Scenes as Compositional Generative Neural Feature Fields"
Python
1,227
star
4

stylegan-t

[ICML'23] StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis
Python
1,122
star
5

transfuser

[PAMI'23] TransFuser: Imitation with Transformer-Based Sensor Fusion for Autonomous Driving; [CVPR'21] Multi-Modal Fusion Transformer for End-to-End Autonomous Driving
Python
957
star
6

stylegan-xl

[SIGGRAPH'22] StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets
Python
939
star
7

projected-gan

[NeurIPS'21] Projected GANs Converge Faster
Python
876
star
8

unimatch

[TPAMI'23] Unifying Flow, Stereo and Depth Estimation
Python
855
star
9

convolutional_occupancy_networks

[ECCV'20] Convolutional Occupancy Networks
Python
792
star
10

differentiable_volumetric_rendering

This repository contains the code for the CVPR 2020 paper "Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision"
Python
782
star
11

mip-splatting

[CVPR'24 Oral] Mip-Splatting: Alias-free 3D Gaussian Splatting
Python
700
star
12

monosdf

[NeurIPS'22] MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction
Python
535
star
13

shape_as_points

[NeurIPS'21] Shape As Points: A Differentiable Poisson Solver
Python
518
star
14

unisurf

[ICCV'21] UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction
Python
410
star
15

graf

Official code release for "GRAF: Generative Radiance Fields for 3D-Aware Image Synthesis"
Jupyter Notebook
393
star
16

tuplan_garage

[CoRL'23] Parting with Misconceptions about Learning-based Vehicle Motion Planning
Python
370
star
17

kitti360Scripts

This repository contains utility scripts for the KITTI-360 dataset.
Python
353
star
18

neat

[ICCV'21] NEAT: Neural Attention Fields for End-to-End Autonomous Driving
Python
291
star
19

gaussian-opacity-fields

Gaussian Opacity Fields for Efficient and Compact Surface Reconstruction in Unbounded Scenes
Python
285
star
20

occupancy_flow

This repository contains the code for the ICCV 2019 paper "Occupancy Flow - 4D Reconstruction by Learning Particle Dynamics"
Python
207
star
21

plant

[CoRL'22] PlanT: Explainable Planning Transformers via Object-Level Representations
Python
192
star
22

factor-fields

[SIGGRAPH 2023] We provide a unified formula for neural fields (Factor Fields) and a novel dictionary factorization (Dictionary Fields)
Jupyter Notebook
183
star
23

carla_garage

[ICCV'23] Hidden Biases of End-to-End Driving Models
Python
121
star
24

texture_fields

This repository contains code for the paper 'Texture Fields: Learning Texture Representations in Function Space'.
Python
113
star
25

sledge

SLEDGE: Synthesizing Simulation Environments for Driving Agents with Generative Models
105
star
26

kitti360LabelTool

JavaScript
103
star
27

counterfactual_generative_networks

[ICLR'21] Counterfactual Generative Networks
Python
102
star
28

gta

[ICLR'24] GTA: A Geometry-Aware Attention Mechanism for Multi-view Transformers
Python
95
star
29

murf

[CVPR'24] MuRF: Multi-Baseline Radiance Fields
Python
84
star
30

controllable_image_synthesis

Towards Unsupervised Learning of Generative Models for 3D Controllable Image Synthesis, CVPR 2020
Python
69
star
31

king

[ECCV'22] KING: Generating Safety-Critical Driving Scenarios for Robust Imitation via Kinematics Gradients
Python
61
star
32

handheld_svbrdf_geometry

On Joint Estimation of Pose, Geometry and svBRDF from a Handheld Scanner, CVPR2020
Python
57
star
33

navsim

NAVSIM: Data-Driven Non-Reactive Autonomous Vehicle Simulation
Python
52
star
34

connecting_the_dots

This repository contains the code for the paper "Connecting the Dots: Learning Representations for Active Monocular Depth Estimation" https://avg.is.tuebingen.mpg.de/publications/riegler2019cvpr
Python
51
star
35

frequency_bias

Official code for "On the Frequency Bias of Generative Models", NeurIPS 2021
Python
39
star
36

data_aggregation

This repository contains the code for the CVPR 2020 paper "Exploring Data Aggregation in Policy Learning for Vision-based Urban Autonomous Driving"
Python
38
star
37

good

[ICLR'23] GOOD: Exploring Geometric Cues for Detecting Objects in an Open World
Python
36
star
38

campari

[3DV'21] CAMPARI: Camera-Aware Decomposed Generative Neural Radiance Fields
Python
29
star
39

autonomousvision.github.io

Blog of the Autonomous Vision Group at MPI-IS Tübingen and University of Tübingen.
HTML
19
star
40

visual_abstractions

6
star
41

slides

Slide repository of the Autonomous Vision Group at MPI-IS Tübingen and University of Tübingen.
CSS
2
star
42

similarity_reconstruction

This code is based on the paper Exploiting Object Similarity in 3D Reconstruction.
C++
1
star
43

slow_flow

This code is based on the paper Slow Flow: Exploiting High-Speed Cameras for Accurate and Diverse Optical Flow Reference Data.
C++
1
star