As an official partner of Swiss company PIX4D, DroneUA continues to integrate advanced photogrammetry, 3D reconstruction, surveying, and digital mapping solutions in Ukraine. The latest PIX4Dmatic update, featuring Gaussian Splatting, expands the capabilities of professional geospatial data processing by combining mobile capture, desktop processing, and cloud collaboration within a single georeferenced workflow.
For DroneUA, this update is significant not just as new software functionality, but as the next step in developing tools that allow specialists to work with 3D data more accurately, consistently, and efficiently — from field data collection to engineering-ready outputs.
Gaussian Splatting is rapidly becoming a key approach to high-precision 3D reconstruction. Its integration into PIX4Dmatic creates a seamless workflow across PIX4Dcatch, PIX4Dcloud, and the desktop environment, providing a unified georeferenced ecosystem for professional spatial data work.
New Approach to Point Cloud Generation
In traditional photogrammetry, the quality of a point cloud depends heavily on densification and image matching accuracy. Complex conditions — such as low-texture surfaces, repeating patterns, or intricate geometry — often result in uneven point density, artifacts, and additional cleanup.
Gaussian Splatting in PIX4Dmatic changes the reconstruction paradigm. Instead of a fragmented model, it generates a more continuous and complete 3D representation, providing:
- More uniform point cloud distribution
- Reduced noise and artifacts
- Improved structural integrity
- A stable foundation for further processing
Point cloud without Gaussian Splatting
Point cloud with Gaussian Splatting. As you can see below, the Gaussian Splatting point cloud is more evenly distributed and has fewer discrepancies.
For professionals, this means not only a visually improved model but also reduced time for data correction and more reliable results.
Higher-Quality Orthomosaics
Orthomosaics are among the most sensitive outputs in any photogrammetric workflow, often revealing weaknesses in reconstruction — especially over water, reflective surfaces, or low-texture areas.
Gaussian Splatting reconstructs scenes as continuous environments, preserving more spatial and visual information. The result is orthomosaics with:
- Greater detail
- Fewer gaps
- Smoother transitions between surfaces
- Reduced data loss
Особливо помітною різниця стає на складних ділянках — зокрема поблизу водних поверхонь, де традиційна фотограмметрія часто демонструє фрагментацію та нестабільність реконструкції.
Standard photogrammetric orthophotoplastomatics in PIX4Dmatic
Orthomosaic generated using Gaussian splicing
The improvement is especially noticeable in challenging areas, such as near water bodies, where traditional photogrammetry often produces fragmented and unstable reconstructions.
For engineering and surveying tasks, this translates directly to more accurate data interpretation, fewer “blind spots,” and more reliable analysis.
More Stable 3D Mesh Generation
3D mesh quality depends on stable input data. Traditional workflows require significant time correcting artifacts, smoothing transitions, and refining triangulation.
With Gaussian Splatting in PIX4Dmatic, meshes are built from a more consistent and aligned dataset, providing:
- More predictable triangulation
- Smoother surface transitions
- Sharper edge detail
- Fewer reconstruction artifacts
3D mesh of Lokovana Bridge
This allows specialists to move directly to analysis and model use without lengthy manual corrections.
Accurate Surface Comparison and Volumetric Analysis
PIX4Dmatic also introduces a new surface comparison tool for volumetric analysis with Gaussian Splatting. It enables volume calculations by directly comparing:
- point cloud;
- TIN surfaces
- Different construction phases
- Survey and as-built models
This analysis is highly sensitive to input data quality. Even minor reconstruction errors can create “false changes,” particularly in:
- Earthwork monitoring
- Stockpile tracking
- Infrastructure inspections
- Construction progress control
Binary visualization mode for surface comparison in PIX4Dmatic
The same dataset with gradient rendering mode for surface comparison in PIX4Dmatic
A more stable point cloud minimizes such errors, ensuring more accurate surface comparisons across datasets.
High-Fidelity Performance
Gaussian Splatting in PIX4Dmatic leverages local GPU resources, maximizing efficiency with modern NVIDIA GPUs with sufficient VRAM.
This approach delivers:
- Faster Gaussian splat generation
- High-resolution scene processing
- Stable desktop processing for large datasets
- Full control over the reconstruction workflow locally
The ecosystem remains flexible — processing can be shifted to PIX4Dcloud without disrupting the unified workflow.
Unified PIX4D Geospatial Ecosystem
A key benefit of this update is integrating Gaussian Splatting across the broader PIX4D ecosystem.
Data captured with PIX4Dcatch can be directly processed in PIX4Dmatic, combining:
- terrestrial scanning;
- aerial photogrammetry;
- RTK workflows;
- mobile mapping;
- cloud collaboration.
Gaussian Splatting в PIX4Dmatic
The result is a unified georeferenced model where terrestrial and aerial data no longer exist separately but are integrated into a single reconstruction pipeline.
PIX4D has confirmed that Gaussian Splatting workflows will soon be available in PIX4Dengine SDK, enabling automated high-volume reconstruction pipelines for large-scale corporate use.
A New Stage in Professional Photogrammetry
The integration of Gaussian Splatting in PIX4Dmatic is more than a new tool—it represents a systematic improvement in geospatial workflows, from mobile capture to engineering-grade analysis.
For surveying, infrastructure monitoring, construction, inspections, and digital mapping, it delivers a new level of reconstruction stability, spatial accuracy, and high-quality deliverables.
