# 3DTopia/OpenLRM

An open-source impl. of Large Reconstruction Models

Repository: https://github.com/3DTopia/OpenLRM
Canonical: https://ross.abutalabs.com/products/openlrm
Language: Python
License: Apache-2.0
License Family: permissive
Topics: 3d, aigc, computer-vision, generation
Last push: 2024-05-06T09:51:00+00:00

## Health v2 (maintenance only)
Score: 17/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 70
- inputs: {"age_days": 987, "days_push": 849, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1244, forks 80 (observed 2026-08-28T04:04:06.918649+00:00)

## What it is
OpenLRM is an open-source PyTorch implementation of Large Reconstruction Models (LRM) that reconstruct 3D objects (meshes and rendered videos) from single images using transformer-based models. It includes pretrained weights trained on Objaverse and MVImgNet, training code, and a Hugging Face demo.

## Use cases
- reconstruct a 3D model from a single image
- generate 3D assets for games or AIGC pipelines
- train a large reconstruction model on custom multi-view data
- run image-to-3D inference with pretrained LRM weights
- render turntable videos of reconstructed 3D objects
- experiment with feed-forward 3D reconstruction research

## When to choose
- you need an open-source, research-grade implementation of LRM with pretrained weights
- you want to fine-tune or train large reconstruction models on your own datasets
- you need single-image to 3D reconstruction with mesh export

## When to avoid
- you need production-ready, photorealistic 3D generation with commercial licensing (weights are CC BY-NC)
- you lack a GPU or cannot install xFormers memory-efficient attention
- you need real-time 3D reconstruction rather than offline feed-forward inference

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, image-processing, graphics
- domain: computer-vision, artificial-intelligence, graphics, deep-learning
- platform: python
- tags: 3d-reconstruction, large-reconstruction-model, lrm, 3d-generation, aigc, objaverse, tripod-gaussian, image-to-3d, gpu, linux

## Member repositories
- 3DTopia/OpenLRM (main) score 17

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:06.918649+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 2026-08-30T05:08:12.102820+00:00, confidence not recorded.
  - readme: https://github.com/3DTopia/OpenLRM (fetched 2026-08-28T04:04:06.918649+00:00, sha 74d056877e6e)
- Data as of 2026-08-30T08:39:29.467469+00:00.