One camera for 2D, two cameras for 3D is many people's first impression of DIC. But today, a single camera can also perform 3D DIC measurement: with only one camera, where does the spatial information come from?

The key is not the number of physical cameras, but whether the system can obtain calibratable, computable different viewing angles. Selection should not start from how many cameras are used, but from how the specimen will deform and what data is needed.

2D DIC: Measuring In-Plane Deformation

2D DICUsually one camera faces the specimen, and in-plane displacement and strain are calculated by tracking the position changes of speckle subsets in the image. Its calibration and deployment are relatively simple, suitable for tension, compression and shear tests where the surface is approximately flat and deformation mainly occurs within the image plane.

But it has an important premise:the displacement of the specimen along the camera optical axis should be small enough.

This is because camera imaging has a perspective effect of closer-is-bigger. When the specimen moves back and forth, the size in the image changes with object distance even without corresponding in-plane deformation. 2D DIC cannot directly distinguish whether this imaging change comes from real in-plane deformation or out-of-plane motion, so it may introduce additional apparent strain.

For tiny strain measurement, the impact of out-of-plane motion deserves particular attention.

Key point: if out-of-plane motions such as bending, warping or fixture yaw may occur during loading, use 2D DIC with caution.

Stereo 3D DIC: reconstructing space through parallax

Stereo 3D DICTwo cameras observe the same speckle area from different angles. The system uses the position difference of the same speckle point in the two views, performs triangulation with calibration parameters, reconstructs the 3D coordinates of the specimen surface, and obtains 3D displacement and surface strain.

The advantage of the stereo approach is more flexible geometric configuration.By adjusting the lens, camera spacing and viewing angle, different fields of view and working distances can be accommodated; for larger-range measurement, it is also convenient to expand through multiple system sets.

What needs to be considered is that the two cameras and acquisition channels require reliable synchronization; on site, the angle, baseline and stereo calibration must be completed, and installation space and sight occlusion issues handled.

Suitable for: curved structures, large components, large field-of-view measurement, and tests involving out-of-plane motion such as bending, torsion and buckling.

Single-camera 3D DIC: two viewpoints from one camera

single-camera 3D DICThrough special optical paths such as prisms and reflective components, two images from different angles are recorded simultaneously on the same sensor. Although it has only one physical camera, it obtains two calibratable viewing viewpoints; matching, calibration and triangulation reconstruction are still required afterwards.

A misconception needs to be clarified here: single-camera 3D is not software-guessed depth from an ordinary photo; it is also different from deep-learning-based monocular depth estimation.In essence, it still belongs to stereo vision measurement.

Its features come from the structure itself: the two viewpoints come from the same exposure, so no trigger synchronization between two cameras is needed; high-speed testing requires only one set of high-speed camera and acquisition channel; the measuring head is more compact, and the integrated optical path relatively simplifies on-site installation and calibration.

It also has clear boundaries: the two viewpoints usually need to share the sensor's effective imaging area, requiring trade-offs among field of view, spatial resolution and measurement range; the equivalent baseline and viewing angles are also less flexible than stereo solutions. For super-large fields of view or large components, stereo 3D is usually more suitable.

Suitable for: high-speed dynamic tests such as Split Hopkinson bar impact and high-frequency vibration; scenarios with limited installation space such as environmental chambers, high-temperature furnaces and highly integrated test setups; and standardized, batch testing tasks with high requirements for configuration consistency, rapid deployment and testing efficiency.

Three Solutions: The Difference Is Not Only the Number of Cameras

The table above is only a basic judgment. For actual selection, four steps are recommended:

▪ First confirm whether only gauge-section data is needed or full-field measurement;

▪ Then judge whether there is out-of-plane motion;

▪ Then look at the field of view and installation space;

▪ Finally, combine testing speed and environmental conditions.

If only axial or transverse strain within the gauge length is of interest, a video extensometer is usually the lighter-weight choice.

Three Common Selection Misconceptions

▪ Using 2D DIC to measure an obviously bending specimen just to save one camera. The problem is not 2D DIC itself, but that out-of-plane motion may be misinterpreted as in-plane deformation.

▪ Believing 3D DIC necessarily requires two physical cameras. What determines 3D measurement capability is the effective viewpoints and spatial geometric relationships, not the number of camera housings.

▪ Comparing only accuracy specs while ignoring site conditions. Synchronization, occlusion, installation space, field of view and calibration method equally determine whether the system can stably obtain valid data.

Haytham has currently formed 2D, stereo 3D, single-camera 3D and single-camera 3D high-speedvisual strain gauge series. The different solutions are not simply high or low configurations, but respectively target different test conditions such as in-plane deformation, conventional 3D, large field of view, high-speed dynamics and confined spaces.

When evaluating a DIC system, the real questions to answer are: what is the measurement target, how will the specimen deform, how large is the required field of view, is the installation space limited, and how fast is the test.

Looking first at data needs and deformation mode, then at field of view, speed and spatial conditions, is closer to correct selection than simply comparing camera numbers.