# facebookresearch/pytorch3d

PyTorch3D is FAIR's library of reusable components for deep learning with 3D data

Repository: https://github.com/facebookresearch/pytorch3d
Canonical: https://ross.abutalabs.com/products/pytorch3d
Homepage: https://pytorch3d.org/
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-08-26T17:50:10+00:00

## Health v2 (maintenance only)
Score: 74/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 27, longevity 100
- inputs: {"age_days": 2505, "days_push": 7, "days_rel": 278, "gap_med": 440, "n_releases_24m": 2}
- 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 9954, forks 1463 (observed 2026-08-28T04:10:38.948145+00:00)

## What it is
PyTorch3D is Facebook AI Research's library of efficient, reusable components for deep learning with 3D data, built on PyTorch. It provides data structures and operators for triangle meshes and point clouds, heterogeneous batching, and a modular differentiable mesh renderer with CUDA acceleration.

## Use cases
- differentiably render 3d meshes in pytorch
- compute chamfer distance between point clouds
- batch 3d meshes of different sizes for deep learning
- train neural networks on triangle mesh data
- new view synthesis with implicit representations
- load and manipulate obj mesh files
- deform 3d meshes with gradient-based optimization

## When to choose
- you need differentiable 3D rendering integrated with PyTorch training loops
- you're doing 3D computer vision research with meshes or point clouds on GPU
- you need efficient 3D operators like graph convolution, sampling, or loss functions on heterogeneous batches

## When to avoid
- you need a general-purpose 3D modeling or game engine tool
- you work only with 2D images or non-PyTorch frameworks like TensorFlow or JAX
- you need a stable high-level API for production rather than research-oriented components

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, graphics, image-processing, math
- domain: deep-learning, computer-vision, graphics, machine-learning
- platform: python, cross-platform
- tags: 3d, pytorch, differentiable-rendering, triangle-meshes, point-clouds, new-view-synthesis, computer-vision-research, cuda, research, gpu, linux, macos

## Member repositories
- facebookresearch/pytorch3d (main) score 74

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:38.948145+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 2026-08-29T17:20:12.674325+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/pytorch3d (fetched 2026-08-28T04:10:38.948145+00:00, sha 56109cdd7be9)
  - homepage: https://pytorch3d.org/ (fetched 2026-08-29T08:20:01.898375+00:00, sha c8bba412736e)
  - site_page: https://pytorch3d.org/docs/why_pytorch3d (fetched 2026-08-29T08:20:01.907353+00:00, sha f0a1d0b1e91f)
  - site_page: https://pytorch3d.org/docs/why_pytorch3d.html (fetched 2026-08-29T08:20:01.909062+00:00, sha f0a1d0b1e91f)
  - registry_pypi: https://pypi.org/pypi/pytorch3d/json (fetched 2026-08-29T08:20:01.910625+00:00, sha 9c74880dbecf)
- Data as of 2026-08-30T08:39:29.467469+00:00.