Skip to content

Point Cloud Tools v2.2.0 - COLMAP 3.14 (GLOMAP Merged)

Choose a tag to compare

View all tags
@lyehe lyehe released this 01 Feb 05:09
· 92 commits to master since this release
v2.2.0
0cbad39

Point Cloud Tools v2.2.0 - COLMAP 3.14

Pre-built Windows/Linux binaries for COLMAP, GLOMAP, and Python wheels with CUDA 12.8 support.

⚠️ Important: NVIDIA Driver 570+ Required

CUDA 12.8 binaries require NVIDIA Driver 570 or later. Check with nvidia-smi.

🎉 Highlights

  • GLOMAP merged into COLMAP - Global SfM functionality is now part of COLMAP 3.14
  • New fisheye (equidistant) camera model
  • Python 3.10-3.14 support

What's Included

COLMAP 3.14 dev - Windows x64 CUDA (119 MB)

File: COLMAP-3.14-dev-Windows-x64-CUDA.zip

Latest COLMAP development version (3.14.0.dev0) for Structure-from-Motion and Multi-View Stereo reconstruction.

  • All dependencies bundled (no separate installation needed)
  • CUDA GPU acceleration for RTX 20/30/40 series, A100, H100
  • Includes GUI and command-line tools

Quick Start:

# Extract and run
.\COLMAP.bat

# Or command line
.\bin\colmap.exe help

GLOMAP - Windows x64 CUDA (667 KB)

File: GLOMAP-Windows-x64-CUDA.zip

Fast global Structure-from-Motion tool.

  • Self-contained with all dependencies (COLMAP 3.11, Ceres, PoseLib)
  • No additional installation required
  • GPU-accelerated processing

Quick Start:

.\bin\glomap.exe mapper --help

pycolmap - Python Wheels (570 MB each)

Files:

  • pycolmap-3.14.0.dev0-cp310-cp310-win_amd64.whl (Python 3.10)
  • pycolmap-3.14.0.dev0-cp311-cp311-win_amd64.whl (Python 3.11)
  • pycolmap-3.14.0.dev0-cp312-cp312-win_amd64.whl (Python 3.12)
  • pycolmap-3.14.0.dev0-cp313-cp313-win_amd64.whl (Python 3.13)
  • pycolmap-3.14.0.dev0-cp314-cp314-win_amd64.whl (Python 3.14)

Python bindings for COLMAP with all dependencies bundled.

  • No separate COLMAP installation needed
  • CUDA support included
  • Works on any Windows machine with matching Python version

Installation:

pip install pycolmap-3.14.0.dev0-cp312-cp312-win_amd64.whl

Quick Test:

import pycolmap
pycolmap.extract_features(image_path="images/", database_path="database.db")

System Requirements

⚠️ Windows Defender False Positive Notice:
Windows Defender may flag these binaries as potentially unwanted software (typically Wacatac.B!ml). This is a false positive common with CUDA-compiled binaries. These files are built from official open-source COLMAP/GLOMAP repositories with no modifications. You can:

Windows Compatibility:

  • ✅ Windows 10/11 (64-bit only) - Fully tested and supported
  • ⚠️ Windows 7/8/8.1 - May work but not tested
  • ❌ 32-bit Windows - Not supported

Minimum:

For GPU Acceleration (optional but recommended):

  • NVIDIA GPU with Compute Capability 7.5+ (RTX 20/30/40, A100, H100)
  • NVIDIA Driver 570.26+ required (for CUDA 12.8 binaries)
    • Windows: Driver 570.65+
    • Linux: Driver 570.26+
  • Check your driver version: nvidia-smi

Note: Binaries will run without CUDA but GPU acceleration will be disabled.

⚠️ "PTX compiled with unsupported toolchain" Error:
If you see this error during GPU feature extraction, your NVIDIA driver is too old.

CUDA error: the provided PTX was compiled with an unsupported toolchain

Fix: Update your NVIDIA driver to version 570 or later:

  • Windows: Download from NVIDIA Drivers
  • Linux: sudo apt install nvidia-driver-570 (Ubuntu/Debian)

For pycolmap:

  • Python 3.10, 3.11, 3.12, 3.13, or 3.14 (64-bit)
  • Wheels are self-contained and work on any compatible Windows machine

What's New in COLMAP 3.14

  • GLOMAP merged into COLMAP - Run global SfM directly: colmap global_mapper
  • New fisheye (equidistant) camera model - Better support for wide-angle lenses
  • Bundle adjustment refactoring - Improved optimization performance
  • PLY mesh reading support - Import mesh files directly
  • Standalone GLOMAP package still available (uses COLMAP 3.11 for compatibility)

Supported GPU Architectures

Compiled for: RTX 20/30/40 series, A100, H100 (architectures 75, 80, 86, 89, 90, 120)

License

Each component has its own BSD 3-Clause License:

Documentation

Built with: Visual Studio 2022, CUDA 12.x, CMake 3.28+, vcpkg

Pre-built COLMAP, GLOMAP, and pycolmap binaries with CUDA support.

Installation

COLMAP

Windows:

Expand-Archive COLMAP-windows-latest-CUDA.zip -DestinationPath C:\Tools\COLMAP
C:\Tools\COLMAP\bin\colmap.exe gui

Linux:

unzip COLMAP-ubuntu-22.04-CUDA.zip -d ~/tools/colmap
~/tools/colmap/bin/colmap gui

GLOMAP

Windows:

Expand-Archive GLOMAP-windows-latest-CUDA12.8.1.zip -DestinationPath C:\Tools\GLOMAP
C:\Tools\GLOMAP\bin\glomap.exe mapper --database_path db.db --image_path images --output_path sparse

Linux:

unzip GLOMAP-ubuntu-22.04-CUDA12.8.1.zip -d ~/tools/glomap
~/tools/glomap/bin/glomap mapper --database_path db.db --image_path images --output_path sparse

pycolmap

Download the appropriate wheel for your Python version and platform, then install:

pip install pycolmap-3.14.0.dev0+cuda-cp312-cp312-win_amd64.whl

Package Variants

Variant Description
CPU CPU-only (no NVIDIA GPU required)
CUDA GPU-accelerated (Windows: self-contained, Linux: requires CUDA Toolkit)
CUDA+cuDSS CUDA + cuDSS sparse solver (faster bundle adjustment)

System Requirements

  • GPU: NVIDIA RTX 20 series or newer (Compute Capability 7.5+)
  • Driver: 520+ (Windows), 525+ (Linux)
  • Python: 3.10, 3.11, 3.12, or 3.13 (for pycolmap)
Assets 40
Loading