Both can create convincing 3D models. The difference is how scale, geometry and uncertainty are established, and that difference becomes critical when the model is used to manufacture something.
Use structured light or laser scanning when metric geometry and repeatable scale are the primary objective, especially on small-to-medium components. Use photogrammetry when the object is large, texture matters, access is difficult or imaging flexibility is more valuable than local dimensional fidelity. Hybrid workflows are often strongest: photogrammetry or targets stabilise global scale while structured light captures local surfaces.
Photogrammetry infers 3D coordinates from features observed across many overlapping photographs. The camera network solves both camera pose and object geometry. Structured-light scanning projects a known pattern or line and triangulates its deformation using calibrated camera/projector geometry. Both are optical triangulation problems, but structured light carries more controlled information into each observation.
That is why photogrammetry can be extraordinarily effective for large scenes with good texture yet become fragile on a smooth, repetitive, glossy or featureless object. Structured light can impose its own features, though it still suffers on reflective, dark or translucent surfaces.
Photogrammetry is scale-ambiguous unless the reconstruction is anchored with known dimensions, scale bars, coded targets or a calibrated multi-camera arrangement. A model can have excellent local shape and still be globally mis-scaled. Structured-light scanners generally inherit metric scale from the calibrated instrument geometry, although tracking and registration can still create accumulated error across a large object.
| Requirement | Photogrammetry | Structured light / laser scanning |
|---|---|---|
| Large-object capture | Excellent with planned image network and scale control | Good, but may require many registrations or a large-field system |
| Texture / colour | Excellent | Varies; geometry is usually the priority |
| Featureless surfaces | Weak unless targets/pattern are added | Stronger because the system projects information |
| Metric scale out of the box | Requires control | Instrument-calibrated |
| Small mechanical details | Possible with macro workflow, but demanding | Usually stronger with the right scanner |
| Low-cost entry | Very strong | Dedicated hardware cost |
The weak version of photogrammetry is “take lots of photos and let the software make a model”. The strong version controls camera calibration, exposure, depth of field, image overlap, convergent angles, scale bars, coded targets, network geometry and independent check dimensions. With that discipline, photogrammetry is extremely useful for vehicles, large castings, site equipment and spatial envelopes.
The method is especially attractive when a scanner cannot physically access the object or when a large number of images can be captured cheaply. Texture can also help document damage, labels, corrosion and assembly context.
A calibrated structured-light system knows the relationship between projector/laser and camera geometry. The operator still has to manage tracking, working distance and surface optics, but the scale does not emerge solely from scene reconstruction. This makes structured light a more natural starting point for reverse engineering a component that will become CAD.
Blue-laser systems add another practical advantage: concentrated laser lines can perform better on some dark or shiny surfaces than broad infrared structured light. That is a workflow advantage, not immunity from optical error.
Current portable scanners increasingly blend multiple acquisition modes. A system may use structured light for local geometry and a photogrammetric target network to stabilise long-range scale. This is sensible engineering: local surface fidelity and global length accuracy are separate problems and can be solved with different sensors.
Revopoint MIRACO Plus is a current example, specifying structured-light capture alongside photogrammetric length measurement. The lesson is broader than the product: do not force one modality to solve every scale of geometry.
Photogrammetry often wins visually because high-resolution photography produces compelling texture maps. That appearance is useful for visualisation and documentation but says little about the geometric error under the texture. For manufacturing decisions, inspect cloud-to-CAD deviation, known lengths and repeat scans.
The papers are useful examples, not universal performance claims; object geometry, scale and procedure strongly affect results.
