Skip to main content

Photogrammetry

TECHNIQUE
📖4 min read

Introduction​

Photogrammetry builds 3D models from overlapping photographs. You capture the subject from many angles, a solver recovers the camera positions and the geometry, and the result lands in Blender for measurement and presentation.

Overlap wins

Shoot with 70 percent or more overlap so the solver finds matches across frames.

One tool chain

COLMAP solves, Blender presents, and the importer moves cameras and clouds between them.

Add scale control

Put one known measurement in the frame set or the reconstruction carries no units.

Key Concepts​

Key Concepts​

  • Photo acquisition
  • Point cloud generation
  • Mesh reconstruction
  • Texture extraction

COLMAP​

COLMAP is an open-source, research-grade photogrammetry pipeline. It's the academic foundation that many other tools build upon. It's highly accurate, well-documented in scientific literature, but:

  • Command-line focused (though there is a GUI)
  • Requires more technical knowledge to use effectively
  • Free and open-source
  • Often used as a preprocessing step or for custom pipelines

Meshroom​

Meshroom is essentially a user-friendly graphical wrapper built on top of the AliceVision photogrammetry framework, which itself draws heavily on COLMAP's methodologies. It's:

  • Free and open-source
  • Node-based visual interface (less intimidating than command-line)
  • Great for learning and experimenting
  • More accessible for beginners

RealityCapture & Metashape​

These are commercial, production-grade tools optimized for speed and handling large datasets:

  • RealityCapture: Known for being extremely fast (GPU-accelerated), popular in game development and surveying
  • Metashape (formerly PhotoScan): Industry standard in archaeology, forensics, and professional surveying; very reliable and well-supported

The relationship: COLMAP pioneered many algorithms (Structure from Motion, Multi-View Stereo) that the others use or build upon. Meshroom democratizes these techniques with a GUI. RealityCapture and Metashape take the core concepts and optimize them for professional workflows with better performance, automation, and support.

For forensics: Many practitioners use RealityCapture or Metashape for the initial 3D reconstruction (speed, reliability), then bring the resulting mesh into Blender for measurement, annotation, and presentation.

Photogrammetry Importer add-on​

The Blender Photogrammetry Importer add-on is specifically designed to import COLMAP reconstruction data directly into Blender.

The Relationship​

COLMAP outputs several file formats during its photogrammetry reconstruction process:

  • Camera positions and orientations (where each photo was taken from)
  • Sparse point clouds (initial 3D feature points)
  • Dense point clouds (more detailed reconstruction)
  • Project files with all this spatial data

The Photogrammetry Importer add-on reads these COLMAP output files and translates them into Blender's native format, allowing you to:

  • Visualize camera positions as objects in 3D space
  • Import the point cloud as geometry
  • Understand the spatial relationships between photos and the reconstructed scene
  • Use the camera data for further analysis or visualization

Forensic Value​

This connection is valuable because:

  1. You can use COLMAP's robust algorithms for the heavy lifting of photogrammetry processing
  2. Then bring the results into Blender for measurement, annotation, and presentation
  3. You get the best of both worlds: COLMAP's accuracy and Blender's visualization/synthesis capabilities

The add-on also supports other photogrammetry software outputs (like Meshroom's AliceVision format), but COLMAP support is central because COLMAP is the open-source standard that many workflows are built around.

In practice: process your crime scene photos in COLMAP, import the reconstruction into Blender with the add-on, add measurements and annotations, then build the presentation.

Workflow​

  1. Acquire: photograph the subject from overlapping angles with fixed exposure.
  2. Solve: run COLMAP or Metashape, then check the sparse point cloud for gaps.
  3. Densify: generate the dense cloud and mesh, and check scale against a known measurement.
  4. Import: bring cameras and clouds into Blender with the Photogrammetry Importer add-on.
  5. Present: measure, annotate, and rebuild what the mesh cannot show.

Best Practices​

  • Keep 70 percent or more overlap between shots.
  • Include a scale bar or a known distance in the frame set.
  • Save the COLMAP or Metashape project files with the case record so the solve repeats.

Common Pitfalls​

  • Textureless surfaces: white walls and open sky give the solver no features to match.
  • Moving objects: anything in motion corrupts the solve. Separate it from the static scene.
  • No scale anchor: import without a known distance and every measurement stays unitless.

Summary​

Photogrammetry turns overlapping photographs into measurable geometry. You solve the cameras, densify the cloud, anchor the scale, and carry the cameras into Blender where the analysis and the presentation happen.