What is photogrammetry?

A measurement method that recovers spatial information from photographs. From many overlapping images of an object, the camera positions are worked out first, and from those the position and shape of the surfaces.

Does siteviewer use photogrammetry?

siteviewer is image-based and shares the starting point with photogrammetry: overlapping footage of the real place. The result, though, is presented as 3D Gaussian Splatting, not as a classic surface model with mesh and texture.

What is the difference between photogrammetry and Gaussian Splatting?

Classic photogrammetry builds explicit geometry: point cloud, mesh and texture, which can be measured and carried into CAD or BIM. Gaussian Splatting builds a representation for rendering new views, which is photorealistic and smooth to move through but does not give you a model of building components.

Which method suits a walkable tour?

For moving freely through a hall with a photorealistic impression, the splatting representation has the edge, because it renders reflections, fine structures and edges without heavy mesh clean-up. If you need geometry you can survey, the route is photogrammetric or scan-based methods.

What footage does an image-based method need?

Enough overlap between shots, taken from different angles, in light that is as even as possible. An ordinary smartphone is enough; for facades and roofs, add a drone.

Are the results accurate enough for surveying?

Not automatically. Measuring is built into the walkthrough, but for a dimension anyone has to stand behind, you still need survey or CAD data produced for that purpose.

Photogrammetry for buildings: method and limits

What photogrammetry is

Photogrammetry is a measurement method that gets spatial information out of photographs. Given many overlapping images of the same object, you can first work out where each picture was taken from, and from those positions the location and shape of the surfaces that appear in them.

The method is far older than digital imaging. Aerial survey was producing maps long before laser scanners existed. What changed is that the calculation now runs automatically and the pictures are allowed to come from a phone.

For buildings and facilities that matters because the instrument drops out. Where someone once had to set up and level a device, a systematic walk with a camera can now be enough.

What photogrammetry produces

Classic photogrammetry produces explicit geometry. The end of the pipeline typically holds:

a point cloud of reconstructed surface points,
a mesh built from it, which is a closed surface,
a texture cut from the original photographs and laid over that mesh,
and, for aerial images, orthophotos and elevation models as well.

The value of these results is that they can be measured and processed further in CAD or BIM tools, provided the capture was planned for it and the accuracy was demonstrated.

Where siteviewer stands

siteviewer is image-based and shares the starting point with photogrammetry: overlapping footage of the real place. You record it yourself, usually as video from a smartphone, and the walkable model is calculated from it.

From there the two part ways. siteviewer presents the scene as 3D Gaussian Splatting: a large set of soft, coloured elements that together produce the picture of the scene from any viewpoint. It is a representation for novel views, not a surface model.

To be explicit: siteviewer is not a classic photogrammetry tool that hands you a surveyed geometry model. The walkthrough is not a mesh you would open in CAD, and it is not a point cloud that someone has checked for accuracy. You can measure in it, and you can download a selected area of a site as points with colour for use elsewhere, but neither turns the walkthrough into a surveyed result. If a survey-grade model is what you need, you need a different tool, and it is better to settle that before the project starts.

Which method fits which job

	Classic photogrammetry	Gaussian Splatting (siteviewer)
Result	Point cloud, mesh, texture	A photorealistic representation to move through freely
Strength	Geometry you can measure; further work in CAD and BIM	Image quality, reflections and fine structures, smooth movement
Weakness	Mesh clean-up is laborious; glass, reflections and thin structures are hard	No model of building components, nothing to hand on to CAD
Suits	As-built survey, planning basis, quantity take-off	Walking the site, coordination, presentation, visual documentation

Both routes can start from the same footage. Which one is right is not decided by the capture but by what the result has to do afterwards.

What the capture needs

Image-based methods make the same demands whichever representation comes out at the end: enough overlap between shots, several angles on the same surface, and light that is as even as you can manage.

In practice a smartphone is enough. For facades and roofs a drone joins in, and for large areas indoors and out an action cam does the job. How such a walk runs in practice is described under 3D building scan.

Large featureless surfaces, strong reflections, people moving through the shot and changing light stay hard, and this holds for both methods alike.

Limits

A siteviewer walkthrough is not automatically a dependable CAD or BIM model. Measuring is part of it, and it covers distance, area, rectangles, circles and volume. For a dimension anyone has to stand behind, you still need survey or CAD data produced for that purpose.

Areas that were not recorded or are hidden cannot be reconstructed, with any image-based method. What the camera did not see is missing from the result, with no sign that it is missing.

How siteviewer builds the reconstruction, and how the approach differs from laser scanning, is covered under product and technology. For what you can do once the model exists, see the 3D walkthrough page, and for how a visual model differs from a data-driven one, digital twin.

Photogrammetry for buildings and facilities