Meshroom has now become a node-based visual programming toolbox for creating, managing, and executing complex data processing pipelines, with a new plugin architecture.
Standard computer vision pipelines such as photogrammetry, camera tracking, HDR panorama, Lidar Meshing, Raw image files conversion and color calibration are now unified within the AliceVision plugin, featuring numerous improvements and optimizations.
Additionally, new AI-powered capabilities include a semantic segmentation plugin and a collection of open-source extensions available via the new MeshroomHub: https://github.com/meshroomHub. This platform enables Gaussian Splatting, monocular depth estimation, and other exploratory features, welcoming developer contributions to expand and enhance these capabilities for upcoming releases.
Highlights
Meshroom New Features
- Advanced Plugin Architecture: Dedicated sub-process isolation for Python nodes with independent local environments
- Integrated Development Tools:
- Built-in Python script editor
- Node’s source code hot-reload for rapid node development iterations
- Enhanced GraphEditor:
- Dynamic output attributes enabling new workflow usages
- New InputNode type enabling interactive evaluation without explicit computation
- Multiple edge disconnection methods and node colorization for better user experience
- Node notifications to attribute changes
- Enhanced 2D Viewer:
- Initial timeline integration with sequence playback controls
- New Reflectance Transformation Imaging (RTI) Viewer: Interactive visualization of albedo and normal maps with real-time lighting control
- New Home Page: featuring pipeline templates and quick access to recent projects.
AliceVision Plugin New Features
- New Pipelines
- Color Calibration: Automated color correction from color charts
- Raw to EXR conversion: Professional image format processing
- Object Reconstruction: Targeted reconstruction with automatic object segmentation
- Turntable Object Reconstruction: Streamlined workflow for rotating object capture
- 360° Object Reconstruction: Reconstruction of complete dual-sided scanning
- LiDAR Processing: Native E57 file import with integrated mesh generation
- Multi-View Photometric Stereo: Advanced surface detail reconstruction with multiple light sources for each viewpoint.
- Pipelines Improvements
- Camera Tracking pipeline: improved stability and reliability
- Introduced experimental fine-grained pipelines for increased modularity and workflow flexibility
- Core Enhancements
- Python Bindings Integration: Enhanced AliceVision accessibility with native Python support for streamlined Machine Learning workflows
New MrSegmentation Plugin
AI segmentation nodes that identify and isolate image objects using natural language prompts, enabling intuitive content-aware processing through foundation models.
MeshroomHub Plugins
We’re excited to introduce new experimental Machine Learning plugins available on MeshroomHub.
These plugins showcase the future of Meshroom workflows, though they currently require developer setup and cannot be installed through the user interface yet.
- mrGSplat: Gaussian Splat optimization and rendering
- mrDepthEstimation: Monocular depth inference
- mrDenseMotion: Optical flow estimation
- mrRoma: Dense deep feature matching
- mrIntrinsicImageDecomposition: Albedo, normals, and material extraction
- mrDeblurring: Video deblurring
- mrGeolocation: GPS extraction and geographic models download
Based on AliceVision 3.3.0.