# nvlabs/neuralangelo

**Attribution required: if you use, quote, or summarise this content, you must credit and link back to [awesome-repositories.com](https://awesome-repositories.com/repository/nvlabs-neuralangelo).**

4,602 stars · 403 forks · Python · NOASSERTION

## Links

- GitHub: https://github.com/NVlabs/neuralangelo
- Homepage: https://research.nvidia.com/labs/dir/neuralangelo/
- awesome-repositories: https://awesome-repositories.com/repository/nvlabs-neuralangelo.md

## Topics

`3d-reconstruction` `computer-graphics` `computer-vision` `neural-rendering` `neural-representation`

## Description

Neuralangelo is a neural surface reconstruction framework that transforms two-dimensional image sequences and multi-view photography into high-fidelity 3D meshes. It implements a pipeline for training neural radiance fields to represent complex scenes as digital geometry.

The project utilizes a signed distance function for surface representation and multi-resolution hash encoding to capture both coarse and fine geometric details. It employs differentiable volume rendering and gradient-based eikonal regularization to ensure the learned distance functions remain physically plausible.

The system includes a full computer vision training pipeline with multi-GPU distributed training and training performance tracking. It provides utilities for photogrammetry data preparation, including video pre-processing, camera pose estimation, and dataset format conversion. A dedicated extraction tool converts trained neural representations into textured 3D meshes using marching cubes and noise removal.

## Tags

### Part of an Awesome List

- [Neural Surface Reconstruction](https://awesome-repositories.com/f/awesome-lists/ai/neural-surface-reconstruction.md) — Implements a neural surface reconstruction framework to transform 2D images into high-fidelity 3D meshes. ([source](https://github.com/NVlabs/neuralangelo/blob/main/README.md))
- [Neural Radiance Field Implementations](https://awesome-repositories.com/f/awesome-lists/ai/neural-radiance-field-implementations.md) — Implements the Neuralangelo algorithm for creating detailed surface reconstructions from multi-view photography.

### Artificial Intelligence & ML

- [Radiance Field Training Pipelines](https://awesome-repositories.com/f/artificial-intelligence-ml/neural-training-pipelines/radiance-field-training-pipelines.md) — Implements a pipeline for optimizing neural networks to represent complex 3D scenes from images.
- [Computer Vision Training](https://awesome-repositories.com/f/artificial-intelligence-ml/computer-vision-training.md) — Provides a GPU-accelerated training workflow for processing image datasets and estimating camera poses.
- [Distributed GPU Training](https://awesome-repositories.com/f/artificial-intelligence-ml/distributed-gpu-training.md) — Supports distributed training across multiple GPUs to accelerate high-resolution surface reconstruction.
- [Eikonal Regularizations](https://awesome-repositories.com/f/artificial-intelligence-ml/gradient-based-learning/eikonal-regularizations.md) — Employs gradient-based eikonal regularization to ensure the learned distance functions remain physically plausible.
- [Neural Training Pipelines](https://awesome-repositories.com/f/artificial-intelligence-ml/neural-training-pipelines.md) — Manages the full training loop and checkpointing across GPUs to optimize 3D surfaces. ([source](https://github.com/NVlabs/neuralangelo/blob/main/train.py))

### Graphics & Multimedia

- [Neural Mesh Extractors](https://awesome-repositories.com/f/graphics-multimedia/3d-mesh-manipulation/neural-mesh-extractors.md) — Provides a pipeline for converting trained neural representations into textured 3D geometry.
- [Differentiable Volume Rendering](https://awesome-repositories.com/f/graphics-multimedia/differentiable-volume-rendering.md) — Uses differentiable volume rendering to optimize the neural surface representation by comparing renders to source images.
- [Surface Representations](https://awesome-repositories.com/f/graphics-multimedia/graphics-engines-rendering/text-rendering/sdf/surface-representations.md) — Represents 3D geometry as a signed distance function where the zero-level set defines the physical surface.
- [Volumetric Mesh Extraction](https://awesome-repositories.com/f/graphics-multimedia/mesh-processing-tools/volumetric-mesh-extraction.md) — Converts trained neural volumetric representations into discrete polygonal triangle meshes using marching cubes. ([source](https://github.com/NVlabs/neuralangelo/blob/main/README.md))
- [Photogrammetry Data Pipelines](https://awesome-repositories.com/f/graphics-multimedia/photogrammetry-data-pipelines.md) — Converts raw video and image sequences into structured formats and camera poses required for reconstruction.
- [Video Frame Extraction](https://awesome-repositories.com/f/graphics-multimedia/video-frame-extraction.md) — Extracts frames from video sequences to generate data for 3D reconstruction. ([source](https://github.com/NVlabs/neuralangelo/blob/main/DATA_PROCESSING.md))

### Software Engineering & Architecture

- [Multiresolution Hash Encoders](https://awesome-repositories.com/f/software-engineering-architecture/hash-tables/multiresolution-hash-encoders.md) — Utilizes multi-resolution hash encoding to capture both coarse and fine geometric details in the scene.