# NVlabs/instant-ngp

Instant neural graphics primitives: lightning fast NeRF and more

Repository: https://github.com/NVlabs/instant-ngp
Canonical: https://ross.abutalabs.com/products/instant-ngp
Homepage: https://nvlabs.github.io/instant-ngp
Language: Cuda
License: NOASSERTION
License Family: other
Topics: neural-network, machine-learning, cuda, nerf, computer-graphics, computer-vision, 3d-reconstruction, signed-distance-functions, function-approximation, real-time, realtime, real-time-rendering
Last push: 2026-02-02T12:32:34+00:00

## Health v2 (maintenance only)
Score: 52/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 65, release rhythm 8, longevity 100
- inputs: {"age_days": 1701, "days_push": 212, "days_rel": 421, "gap_med": null, "n_releases_24m": 1}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 17535, forks 2066 (observed 2026-08-28T04:11:19.966904+00:00)

## What it is
NVIDIA's implementation of instant neural graphics primitives, training NeRFs, signed distance functions, neural images, and neural volumes in seconds using multiresolution hash encoding on a single GPU. It ships as a desktop application with prebuilt Windows binaries and Python bindings built on the tiny-cuda-nn framework.

## Use cases
- train a NeRF from photos in seconds
- reconstruct 3D scenes from images
- learn signed distance functions of 3D surfaces
- represent gigapixel images as neural networks
- render captured scenes in real time
- denoise volumetric radiance fields from path tracers

## When to choose
- you need extremely fast NeRF or SDF training on an NVIDIA GPU
- you want real-time rendering of neural radiance fields
- you want a reference implementation of multiresolution hash encoding
- you want prebuilt binaries to try NeRFs without building code

## When to avoid
- you don't have a CUDA-capable NVIDIA GPU
- you need permissively licensed code for commercial products (research license only)
- you need CPU-only training
- you need a production 3D reconstruction pipeline rather than a research tool

## Facets
- artifact type: application
- maturity: stable
- function: machine-learning, computer-vision, graphics, image-processing, simulation
- domain: computer-vision, graphics, machine-learning, deep-learning
- platform: windows, python
- tags: nerf, neural-rendering, 3d-reconstruction, signed-distance-functions, hash-encoding, tiny-cuda-nn, real-time-rendering, neural-volumes, gigapixel-images, linux, cuda, gpu

## Member repositories
- NVlabs/instant-ngp (main) score 52

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:19.966904+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:02:54.196788+00:00, confidence not recorded.
  - readme: https://github.com/NVlabs/instant-ngp (fetched 2026-08-28T04:11:19.966904+00:00, sha 2dddfdd4e984)
  - homepage: https://nvlabs.github.io/instant-ngp (fetched 2026-08-29T08:00:53.650124+00:00, sha 3964b4cfdbe4)
- Data as of 2026-08-30T08:39:29.467469+00:00.