3D scanning and ASCAND glossary
3D scanning connects physical capture, optical evidence, computation, geometry and downstream use. A shared vocabulary matters because familiar words such as scan, model, accuracy and resolution can refer to different things at different stages.
This glossary defines how important terms are used across the ASCAND website. Each definition is intentionally bounded. It explains a concept; it does not establish a product specification, guarantee a result or replace the current instructions for a particular workflow.
When a deeper article is available, the definition links to the page that owns the complete explanation.
Start with what 3D scanning means
Find a term
Use your browser’s page search. Terms are grouped by their role so that related concepts remain visible together.
Preferred term identifies the wording used consistently across this site. Also called records a useful alias. Not the same as marks a boundary that prevents two related concepts from being treated as interchangeable.
Acquisition and orientation
Acquisition The stage in which a system records observations of a physical subject. Acquisition includes the camera or sensor relationship, movement, framing, lighting, background, visibility and sequence—not only pressing Record. Calibration Establishing parameters that relate measurements or image observations to a geometric model. Depending on the system, these can describe a camera, lens, laser plane, scale or relationships among components. A calibrated setup is not automatically accurate under every condition. Capture The practical act and resulting data recording through which observations are acquired. On this site, capture precedes reconstruction. Not the same as: a finished 3D model. Coordinate frame A defined origin and set of axes used to express positions and orientations. Separate captures can describe geometry in different coordinate frames until registration relates them. Frame One image in an ordered video or image sequence. A frame can contain object appearance, silhouette, coded orientation or laser-line evidence, but it is not itself a three-dimensional measurement of every visible pixel. Gray Code A coding scheme in which neighboring values are arranged to differ in a controlled way. ASCAND’s visible coded carrier contributes orientation-related observations during rotation. It does not measure object geometry or certify angular accuracy. [Learn about structured rotation](/knowledge/structured-rotation-gray-code/#why-gray-code). Handheld workflow An acquisition relationship in which an operator moves a camera or sensor around a subject. The term describes movement, not one sensing technology, accuracy class or reconstruction method. Orientation reference Evidence used to associate an observation with the object’s rotational state or another spatial orientation. Not the same as: object-surface evidence, complete calibration or proof that a reconstruction is correct. Relative motion A change in the spatial relationship between camera and object. It can be produced by moving the camera, rotating the object or both. ASCAND creates changing views while the camera remains physically steady. [See why the camera stays still](/knowledge/steady-camera-rotating-object/#relative-motion). Structured rotation Controlled object rotation that produces an ordered set of viewpoints and orientation context. The structure constrains acquisition; it does not guarantee visibility, decoding, accuracy or completeness. Turntable workflow An acquisition relationship in which an object rotates while a camera or sensor observes it from a controlled position. Not the same as: one particular reconstruction method. Viewpoint The position and orientation from which a subject is observed. A sequence of different relative viewpoints supplies evidence about different visible portions of the object.
Evidence and observation
Camera ray A conceptual line extending from the camera center through an image location into the scene. In laser triangulation, intersecting a camera ray with the modeled laser plane can yield a three-dimensional surface sample. Direct evidence An observation that directly constrains a reconstruction under the selected method—for example, an observed silhouette boundary or detected laser response. Direct does not mean error-free or complete. Evidence Recorded information used to constrain, support or evaluate a reconstruction. Different evidence types answer different geometric questions and should not be treated as interchangeable. Feature correspondence An estimated match between recognizable image features in different views. Conventional photogrammetric processing can use correspondences to estimate camera relationships and reconstruct points. ASCAND should not be reduced to this one principle. [Compare ASCAND and photogrammetry](/knowledge/ascand-and-photogrammetry/#shared-principles). Laser plane The geometric plane of light projected by a line laser under the system model. The visible line on the object is the surface intersection with that plane—not the plane itself. Laser triangulation A method that derives surface samples by relating a detected laser response, a camera ray and a modeled laser plane. Both the laser and camera need suitable visibility. [Read the full principle](/knowledge/laser-triangulation/#triangulation-principle). Line of sight An unobstructed optical path between the observing system and a surface. Active methods can require two paths: illumination must reach the surface and the camera must observe its response. Material response How a surface absorbs, reflects, transmits or scatters the light used during capture. Dark, glossy, reflective, translucent or transparent behavior can affect evidence differently depending on method and setup. Observation Recorded information about the subject or capture relationship. An observation becomes useful geometric evidence only through the assumptions and processing of a particular method. Occlusion A condition in which one surface blocks another from a camera, sensor or illumination source. Rotation can change occlusion, but no movement pattern automatically reveals every surface. Segmentation Separating image regions into categories such as foreground object and background. In silhouette reconstruction, segmentation quality directly affects the extracted outline and resulting volume constraints. Silhouette The two-dimensional outline of the object as observed from one viewpoint. It constrains which projected regions may be occupied but does not directly reveal depth or hidden concavities. Visual cone The three-dimensional volume extending from the camera through an observed silhouette. Intersecting constraints from multiple visual cones produces a visual hull.
Reconstruction, alignment and combination
Coarse alignment An initial estimate that brings separate datasets into approximately corresponding positions and orientations before local refinement. Combo scanning An ASCAND workflow that coordinates complementary evidence paths while preserving their separate reconstruction responsibilities. Related capture does not automatically fuse the outputs or make them mutually corrective. [Understand Combo evidence](/knowledge/combo-scanning/#complementary-evidence). Fine registration Local refinement of an existing alignment using overlapping geometry or other supported correspondences. It needs a sufficiently plausible starting relationship and does not prove that the final alignment is globally correct. Fusion Combining information from different evidence sources or representations under defined rules. Not the same as: merely placing two datasets in the same coordinate frame or taking their unexamined union. ICPIterative Closest Point. A family of algorithms that iteratively refines alignment between overlapping geometric datasets using proximity-based correspondences. ICP can converge to an incorrect local alignment, especially with weak overlap, symmetry or a poor initial pose. Merge Combining registered datasets or reconstructions into a cumulative result. A merge can retain noise, seams, conflicts and gaps; it is not automatically a finished surface. Multi Merge The downstream registration and combination of multiple reconstructions. Not the same as: Multi-Scan acquisition or Combo evidence reconciliation. Multi-Scan An acquisition strategy that captures the same object in separate orientations or passes to obtain complementary observations. Useful additional coverage still needs enough reliable overlap or another documented alignment basis. [Read about Multi-Scan and Multi Merge](/knowledge/multi-scan-multi-merge/#complementary-orientations). Point sample A three-dimensional position derived or estimated from the available evidence. A sample describes a location, not a continuous surface area. Reconstruction The computational process of estimating three-dimensional geometry from recorded observations and method assumptions. A reconstruction is a representation derived from evidence—not the physical object itself. Registration Estimating a transformation that brings separate datasets into a shared coordinate relationship. Successful numerical convergence is not, by itself, proof of correct alignment. Rigid transformation A rotation and translation that changes a dataset’s pose without changing its internal shape or scale. Visual hull The largest volume consistent with a set of observed silhouettes under the reconstruction model. It can approximate the exterior outline well while filling concavities that never affect a silhouette. [Learn about the visual hull](/knowledge/vision-scanning-voxel-carving/#visual-hull). Voxel A discrete cell in a three-dimensional grid. Voxels can represent candidate occupied or empty regions during volumetric reconstruction. Voxel carving A reconstruction approach that begins with a bounded candidate volume and removes voxels contradicted by observations. In silhouette carving, surviving voxels approximate the visual hull.
Geometry, representations and files
CAD model A model created for computer-aided design, often using explicit dimensions, constraints, features or parametric relationships. A scan-derived mesh is not automatically editable CAD or recovered design intent. Digital twin A digital representation connected to a defined physical asset, context and purpose, often with maintained data or operational relationships. A standalone 3D scan should not automatically be called a complete digital twin. Edge In a polygon mesh, a connection between two vertices that helps bound one or more faces. Face A polygonal element bounded by edges in a mesh. Triangular faces are common, but the presence of faces does not establish surface correctness. File format A defined structure for storing and exchanging data. Format determines which data can be represented—not whether the geometry is accurate, complete or suitable for its intended use. Mesh A surface representation formed by vertices, edges and faces. A mesh may be open, closed, textured, noisy, repaired or inferred. [Compare point clouds and meshes](/knowledge/point-clouds-and-meshes/#geometry-stages). Point cloud A collection of three-dimensional sample positions, sometimes with color, normals or confidence-related attributes. A point cloud does not explicitly connect samples into a continuous surface. Polygon A planar face bounded by edges. In 3D meshes, polygons approximate a surface at the chosen sampling and meshing resolution. Post-processing Operations performed after reconstruction, such as filtering, registration, cropping, hole treatment, remeshing, simplification, scaling, texturing or preparation for a downstream tool. Each operation can change the representation and should be evaluated. Surface normal A direction perpendicular to a local surface or associated with a point or face. Normals support shading, orientation and some processing operations; incorrect normals can make otherwise present geometry render incorrectly. Texture Image-based appearance mapped onto geometry using coordinates. Texture can make a model look detailed without adding geometric detail. Topology The connectivity and organization of a mesh’s vertices, edges and faces. Good visual appearance does not necessarily imply suitable topology for editing, animation, analysis or fabrication. Vertex A position used as a fundamental element of a mesh. Edges connect vertices and faces are bounded by those edges. Vertex color Color values stored on mesh vertices and interpolated across faces. Not the same as: image texture or additional surface geometry. Watertight mesh A closed mesh without open boundary edges under the relevant topology test. Watertightness does not prove that the enclosed shape is accurate, directly observed, printable or fit for use.
Quality, uncertainty and intended use
Accuracy Closeness of a result to an accepted reference value or geometry under a defined test. Accuracy requires a reference, scope and method; it should not be inferred from visual smoothness or file resolution. Artifact A feature in the result caused by acquisition, processing or representation rather than the physical subject—for example, false surfaces, streaks, ripples, seams or duplicated geometry. Completeness The extent to which required regions are represented. A model can be topologically closed yet incomplete or incorrect relative to the physical object and intended use. Coverage The extent of the subject for which usable observations or reconstruction data exist. Coverage should be evaluated against required surfaces, not only total visible area. Fitness for purpose Whether a result is adequate for a stated downstream use and its acceptance criteria. The same reconstruction may be suitable for visualization but unsuitable for dimensional inspection or interface design. Inferred geometry Geometry created through modeling, interpolation, completion or assumptions without equivalent direct evidence for that region. It can be useful but should not be presented as directly observed. Interpolation Estimating values between available samples under a mathematical rule. Interpolation can create continuity; it does not create new observation evidence. Noise Unwanted variation in observations or geometry that obscures the underlying signal or surface. Filtering may reduce noise but can also remove real detail. Outlier A sample that is inconsistent with the main local or global data pattern. Outliers may arise from detection errors, reflections, background inclusion, misalignment or other causes. Precision Repeatability or closeness among repeated measurements or estimates. A precise result can still be systematically inaccurate. Resolution The sampling or representation granularity available at a defined stage, such as image pixels, voxel size, point spacing or mesh density. Higher nominal resolution does not guarantee higher accuracy or more truthful detail. Scale The relationship between model units and physical dimensions. Correct-looking proportions do not establish correct physical scale. Scan A capture-and-reconstruction process or its resulting digital representation, depending on context. The word alone does not promise a complete copy, measured CAD model, certified dimensions or production-ready asset. Scan quality A context-dependent judgment across several dimensions, including evidence, coverage, geometric consistency, accuracy where tested, appearance and fitness for purpose. It should not be reduced to one universal score. Tolerance An allowed variation from a specified nominal value or requirement. Tolerance belongs to a design, manufacturing or acceptance context; it is not automatically contained in scan geometry. Validation Checking a result against defined requirements, references or intended-use criteria. Visual inspection is useful but may not be sufficient when dimensions, fit, safety or regulated decisions matter.
Use the five-part scan evaluation framework
A definition is the beginning of the explanation
The glossary keeps terminology consistent, but different questions belong to different parts of the website:
- Knowledge articles explain concepts, relationships and limitations.
- Tutorials contain current, task-specific operating steps.
- Support pages diagnose problems and route users to verified remedies.
- Product pages own current contents, requirements, compatibility and commercial facts.
- Gallery items provide bounded evidence about named objects, methods and outcomes.
If a term is used differently in another technical field, read it here within the stated 3D scanning and ASCAND context. If a definition appears to conflict with current product instructions, follow the verified product or tutorial source for the operating decision and flag the terminology for editorial review.
Review verified examplesTurntable and handheld scanning differ first in how they create changing views of a subject.