# DLR-RM/BlenderProc

A procedural Blender pipeline for photorealistic training image generation

Repository: https://github.com/DLR-RM/BlenderProc
Canonical: https://ross.abutalabs.com/products/blenderproc
Language: Python
License: GPL-3.0
License Family: copyleft
Topics: blender-pipeline, segmentation, depth-images, camera-positions, suncg-scene, camera-sampling, blender-installation, synthetic, blender, rendering, pose-estimation, synthetic-data, python, 3d-graphics, computer-graphics, 3d-reconstruction, 3d-engines, 3d-front-dataset
Last push: 2026-01-20T18:01:09+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 63, release rhythm 40, longevity 100
- inputs: {"age_days": 2519, "days_push": 225, "days_rel": 680, "gap_med": 0, "n_releases_24m": 2}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3684, forks 518 (observed 2026-08-28T04:08:13.836613+00:00)

## What it is
BlenderProc is a procedural Python pipeline built on Blender for generating photorealistic synthetic training images with ground-truth annotations. It automates scene loading, object pose sampling, lighting, camera placement, and rendering of RGB, depth, normal, and segmentation outputs.

## Use cases
- generate synthetic training images for object detection
- render depth and segmentation ground truth for pose estimation
- create photorealistic datasets from 3D models like ShapeNet or 3D-FRONT
- sample camera poses in a 3D scene for computer vision training
- produce COCO or BOP annotated synthetic data
- simulate physics and collisions in generated scenes

## When to choose
- you need labeled synthetic data for training computer vision models
- you want reproducible, scriptable Blender-based rendering pipelines
- you need RGB, depth, normal, and segmentation outputs from the same scene
- you work with standard 3D asset datasets like BOP, ShapeNet, Haven, or 3D-FRONT

## When to avoid
- you need real photographs rather than synthetic renders
- you want a GUI-driven 3D modeling workflow instead of scripted pipelines
- your project cannot accept GPL-3.0 licensing constraints
- you need lightweight rendering without a full Blender installation

## Facets
- artifact type: library
- maturity: active
- function: image-processing, computer-vision, machine-learning, data-generation, simulation, graphics, cli
- domain: computer-vision, machine-learning, graphics, simulation, artificial-intelligence
- platform: python, cli
- tags: blender, synthetic-data, photorealistic-rendering, segmentation-masks, depth-images, camera-sampling, 3d-scene-generation, training-data, linux, macos

## Member repositories
- DLR-RM/BlenderProc (main) score 62

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:13.836613+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 2026-08-29T18:31:11.010707+00:00, confidence not recorded.
  - readme: https://github.com/DLR-RM/BlenderProc (fetched 2026-08-28T04:08:13.836613+00:00, sha 89b4db4c714e)
  - registry_pypi: https://pypi.org/pypi/blenderproc/json (fetched 2026-08-29T09:24:56.474864+00:00, sha dad7af2c8006)
- Data as of 2026-08-30T08:39:29.467469+00:00.