# NVlabs/nvdiffrec

Official code for the CVPR 2022 (oral) paper "Extracting Triangular 3D Models, Materials, and Lighting From Images".

Repository: https://github.com/NVlabs/nvdiffrec
Canonical: https://ross.abutalabs.com/products/nvdiffrec
Language: Python
License: NOASSERTION
License Family: other
Topics: pytorch, deep-learning
Last push: 2026-08-12T06:42:47+00:00

## Health v2 (maintenance only)
Score: 76/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 97, release rhythm 35, longevity 100
- inputs: {"age_days": 1743, "days_push": 21, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2296, forks 252 (observed 2026-08-28T04:06:35.178046+00:00)

## What it is
nvdiffrec is NVIDIA's official implementation of a CVPR 2022 oral paper that jointly optimizes triangular 3D meshes, PBR materials, and lighting from multi-view images using differentiable rendering and marching tetrahedra. It is a research codebase built on PyTorch with CUDA extensions for differentiable rasterization.

## Use cases
- reconstruct a textured 3D mesh from multi-view photos
- extract PBR materials and lighting from images of an object
- inverse rendering research experiments
- generate triangle meshes with topology optimization from image observations
- compare differentiable isosurfacing techniques like FlexiCubes
- reproduce CVPR 2022 paper results

## When to choose
- you need research-grade inverse rendering with mesh, material, and lighting joint optimization
- you want to reproduce or extend the nvdiffrec paper
- you have a CUDA-capable GPU and multi-view image datasets
- you want to experiment with differentiable marching tetrahedra or FlexiCubes

## When to avoid
- you need a production-ready photogrammetry tool with a GUI
- you lack a CUDA GPU or cannot build CUDA extensions
- you need a permissively licensed library (it uses the Nvidia Source Code License)
- you only need simple 3D scanning without material/lighting estimation

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning, image-processing, graphics, simulation
- domain: computer-vision, graphics, machine-learning, deep-learning
- platform: python, windows, cross-platform
- tags: 3d-reconstruction, inverse-rendering, differentiable-rendering, pytorch, mesh-extraction, pbr-materials, research-code, nvidia, gpu, linux

## Member repositories
- NVlabs/nvdiffrec (main) score 76

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:35.178046+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 2026-08-30T02:40:19.632476+00:00, confidence not recorded.
  - readme: https://github.com/NVlabs/nvdiffrec (fetched 2026-08-28T04:06:35.178046+00:00, sha 118c493845fe)
- Data as of 2026-08-30T08:39:29.467469+00:00.