Both produce a point cloud. How each one gets there — inferring geometry from photos vs. measuring distance directly — is exactly why they don't perform the same on the same surface.
Photogrammetry and LiDAR both end up producing the same kind of output — a 3D point cloud of real-world surfaces — which makes it tempting to treat them as interchangeable, differing mainly in equipment cost. They aren't interchangeable, because they get to that point cloud through fundamentally different physics. Photogrammetry infers 3D positions by triangulating matched features across many overlapping 2D photographs. LiDAR measures 3D positions directly, by timing how long a laser pulse takes to bounce off a surface and return. That difference in method is exactly what explains where each one is accurate, and where each one struggles — this is a capture-method comparison, distinct from what a point cloud becomes after capture (see the Point Cloud vs. BIM Model explainer for that next step).
Photogrammetry captures dozens to thousands of overlapping standard photographs from many angles, then uses Structure-from-Motion (SfM) software to find the same distinctive visual features across multiple photos and triangulate where each one must sit in 3D space. It needs good, consistent lighting and enough visual texture for the software to reliably match the same feature between photos. LiDAR (Light Detection and Ranging) fires a laser pulse at a surface and calculates distance directly from the time (or phase shift) of the return signal — no feature matching involved, and no dependence on ambient light, since the scanner supplies its own light source.
Photogrammetry never measures distance at all — it infers 3D position purely by finding the same visual feature in multiple photos and triangulating. That means its accuracy is entirely dependent on there being enough visual texture and consistent lighting for software to confidently say "this pixel in photo 1 and this pixel in photo 7 are the same physical point." A blank painted wall, a repetitive tile pattern, or a glossy reflective surface gives the algorithm nothing reliable to match — accuracy degrades or the reconstruction fails outright. LiDAR sidesteps that problem entirely, because it never needs to recognize a feature — it just times a pulse of light it fired itself. That's why LiDAR keeps working in a pitch-dark mechanical room or on a blank concrete wall where photogrammetry would struggle. The tradeoff runs the other way for transparent and highly reflective surfaces: a laser pulse can pass straight through glass or bounce unpredictably off a mirror instead of returning a clean, single reflection, producing noisy or missing points — a failure mode photogrammetry doesn't share in quite the same way, since a camera can still often photograph the visible frame or texture around glazing.
Cost is a real factor — commodity cameras and drones are dramatically cheaper than survey-grade terrestrial or mobile LiDAR scanners — but treating that as the only difference undersells a genuine accuracy and reliability gap. LiDAR is generally the more accurate and more consistent choice for hard-surface, low-texture as-built capture, because it measures distance directly rather than inferring it — terrestrial LiDAR scanners routinely hit millimeter-level accuracy independent of lighting or surface pattern. Photogrammetry can produce excellent results too, and it captures true color and texture far more naturally and cheaply than LiDAR does, but its accuracy is conditional: it can genuinely struggle or fail outright on reflective, transparent, or highly repetitive surfaces where the underlying feature-matching has nothing distinct to lock onto. Neither method is strictly "better" — many existing-conditions capture workflows deliberately combine both, using LiDAR for dimensional accuracy on structure and hard surfaces, and photogrammetry (or LiDAR-mounted cameras) for color and texture, rather than treating the choice as a single either/or budget decision.
Explains the real difference between photogrammetry and LiDAR as reality-capture methods — triangulating matched features across overlapping photos versus measuring distance directly with a laser pulse — and why that mechanism, not just equipment price, explains where each one is more accurate and where each one struggles.
Both methods produce the same kind of output — a 3D point cloud — and both get described as "3D scanning" in casual conversation, which hides a real mechanical difference underneath. Photogrammetry never directly measures a distance; it takes many overlapping ordinary photographs and uses Structure-from-Motion (SfM) algorithms to identify the same visual feature across multiple images and triangulate its 3D position, the same basic principle as human stereo vision but across dozens or hundreds of viewpoints. LiDAR (Light Detection and Ranging) measures distance directly, timing how long an emitted laser pulse takes to travel to a surface and return (time-of-flight), or measuring the phase shift of a continuous laser beam — no visual feature matching involved at all.
Because photogrammetry depends on matching visual features, its accuracy is conditional on the scene actually having distinct, consistently-lit visual detail — a textured brick facade in daylight is an easy case; a blank white wall, a repetitive facade pattern, or glossy/reflective surfaces (glass, polished metal, chrome) are hard cases where feature matching becomes unreliable or fails outright. LiDAR sidesteps that dependency because it never needs to recognize anything — it just times its own emitted pulse — so it performs consistently on textureless, low-light, or visually repetitive hard surfaces where photogrammetry struggles. LiDAR has its own weak points, though: highly reflective or transparent surfaces (glass, mirrors, polished chrome) can cause the laser to pass through or bounce unpredictably rather than return a single clean reflection, producing noise or gaps in the resulting point cloud.
On real AEC projects, this isn't purely an accuracy debate — it drives actual equipment and workflow decisions. Terrestrial or mobile LiDAR scanning is the standard choice for high-accuracy, survey-grade as-built capture of structure, MEP, and hard interior/exterior surfaces, particularly on renovation projects where dimensional accuracy against real conditions is critical. Photogrammetry (often drone-based for building exteriors and large sites) is commonly chosen for its lower equipment cost, faster field deployment, and its natural strength at capturing true color and surface texture — useful for visualization, historic documentation, and site context. Many capture workflows now combine both, using LiDAR-scanned geometry as the dimensional backbone and photogrammetry-derived or camera-captured texture as the visual layer on top.
Generally yes for hard-surface, low-texture as-built capture, because it measures distance directly rather than inferring it from matched photo features — terrestrial LiDAR routinely achieves millimeter-level accuracy independent of lighting or surface pattern. But photogrammetry can achieve very high accuracy too under good conditions (consistent lighting, sufficient texture, enough photo overlap), and it captures true color/texture more naturally, so "always more accurate" oversimplifies a real tradeoff rather than a strict hierarchy.
Because those surfaces don't present a consistent, matchable visual feature between photos — a reflection changes depending on camera angle and lighting, so the same physical point on the glass looks different from each camera position, breaking the feature-matching step Structure-from-Motion relies on. LiDAR also struggles with glass and mirrors, but for a different reason: the laser pulse can pass through transparent material or reflect off a mirror at an angle that never returns to the scanner.
Not inherently — a LiDAR scanner's core output is geometric distance measurements, which is why raw LiDAR point clouds are often shown in grayscale or intensity-based coloring. Most commercial terrestrial LiDAR scanners solve this by including a built-in camera that captures color photos alongside the scan and maps that color onto the point cloud afterward, but the color and the distance measurement come from two different sensors, not the same laser measurement.
Photogrammetry is typically far cheaper to get started with, since it only requires a good camera or drone rather than dedicated laser scanning hardware, which can cost tens of thousands of dollars for survey-grade units. That cost gap is real and often decisive for smaller projects — the tradeoff is accepting photogrammetry's greater sensitivity to lighting and surface texture conditions.
Yes, and it's an increasingly common practice — using LiDAR to establish accurate dimensional geometry on structure and hard surfaces, and photogrammetry (or a scanner's onboard camera) to supply realistic color and texture, or using photogrammetry to fill in areas outside a LiDAR scanner's practical range or line of sight. The two point clouds are typically registered into the same coordinate system so they can be used together as one unified reality-capture dataset.
Try our BIM, CAD & Digital Design Studio
More calculators, simulators, and guides for this discipline.