# MeshAnything

[ICCV 2025] From anything to mesh like human artists. Official impl. of "MeshAnything V2: Artist-Created Mesh Generation With Adjacent Mesh Tokenization"

Repository: https://github.com/buaacyw/MeshAnythingV2
Canonical: https://ross.abutalabs.com/products/meshanything
Homepage: https://buaacyw.github.io/meshanything-v2/
Language: Python
License: NOASSERTION
License Family: other
Topics: 3d, 3d-generation, generative-model, mesh, transformers, iccv2025
Last push: 2025-04-28T16:36:51+00:00
Link (homepage): https://buaacyw.github.io/meshanything-v2/

## Health v2 (maintenance only)
Score: 31/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 18, release rhythm 35, longevity 54
- inputs: {"age_days": 758, "days_push": 492, "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 1015, forks 73 (observed 2026-08-28T04:03:14.047878+00:00)

## What it is
MeshAnything is an autoregressive transformer model that generates artist-created 3D meshes (up to 1600 faces in V2) aligned with a given shape, such as a point cloud or 3D asset. It integrates into 3D asset production pipelines to produce compact, human-artist-like meshes instead of dense reconstructions.

## Use cases
- convert a point cloud or 3D shape into a clean artist-style mesh
- generate low-poly game-ready meshes from 3D assets
- post-process 3D generation pipeline outputs into editable meshes
- research autoregressive mesh tokenization
- create controllable 3D meshes aligned to a reference shape

## When to choose
- you need compact, artist-like meshes rather than dense triangle soups
- you have a GPU (A100/A6000-class) and want state-of-the-art mesh generation
- you want to integrate mesh generation into an existing 3D asset pipeline

## When to avoid
- you need meshes with more than ~1600 faces
- you lack a CUDA GPU with significant VRAM
- you need permissively licensed code for commercial use (SLab license)
- you need production-ready textured assets, not geometry only

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-inference, graphics, image-processing
- domain: deep-learning, large-language-models, graphics, computer-vision, artificial-intelligence
- platform: python, cli
- tags: mesh-generation, autoregressive-transformer, 3d-assets, artist-created-mesh, iclr2025, point-cloud-conditioned, non-commercial-license, linux, gpu

## Member repositories
- buaacyw/MeshAnythingV2 (main) score 31
- buaacyw/MeshAnything (mirror) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:14.047878+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 2026-08-30T02:40:14.098791+00:00, confidence not recorded.
  - readme: https://github.com/buaacyw/MeshAnythingV2 (fetched 2026-08-28T04:03:14.047878+00:00, sha 42ed6b31a17c)
  - homepage: https://buaacyw.github.io/meshanything-v2/ (fetched 2026-08-29T10:20:32.135090+00:00, sha 128995b02f6b)
- Data as of 2026-08-30T08:39:29.467469+00:00.