Understand what happens between a physical object and a digital model
3D scanning is a family of processes that uses observations of a physical object or environment to create a digital representation of its geometry.
That definition has four distinct parts:
- A physical subject exists in the real world.
- A capture method records evidence about it.
- A reconstruction process converts or combines that evidence into spatial geometry.
- The geometry is stored in a digital representation, such as a point cloud or mesh.
The word scanning can make this sound like a direct copy operation. In practice, every method observes the subject through a particular physical relationship. A camera records visible light. A laser system observes how a projected line appears on a surface. A depth sensor estimates distance. A contact probe records selected positions by touching them.
The resulting model therefore represents what the method could observe, measure or constrain—and what the reconstruction could derive from that evidence. It is not automatically a complete digital twin, an editable CAD model, a certified measurement record or a print-ready object.
See how evidence becomes geometry
Explore the 3D Scanning Knowledge Center
A scan begins with evidence, not with a finished model
Although scanning systems differ, the conceptual path from object to model usually contains five parts.
1. Establish a spatial relationship
The system needs a way to relate observations to positions or directions in space. That relationship may come from calibrated cameras, known sensor geometry, controlled motion, coded references, tracked movement or direct contact coordinates.
Without a usable spatial relationship, separate observations cannot be placed consistently into one model.
2. Observe the subject
The capture method records evidence. Depending on the method, this may include photographs, silhouettes, projected patterns, a laser line, distance samples or contact points.
An observation is not yet the entire surface. It records only what the capture arrangement and physical method can detect.
3. Locate or constrain geometry
The system interprets the observations geometrically. Some methods calculate surface positions directly from measured relationships. Others constrain where the surface can or cannot be. For example, a silhouette can rule out volumes outside the observed outline without describing every hidden concavity.
4. Combine evidence
Observations from different viewpoints, rotation angles, sensors or object orientations are brought into a shared coordinate system. Agreement can strengthen the reconstruction. Missing, inconsistent or poorly observed evidence can create gaps, noise or ambiguity.
5. Represent the result
The reconstructed geometry may first appear as points, samples, voxels or another intermediate form. Later processing may connect or convert it into surfaces and output files.
This chain explains why capture conditions matter. If an important region was never visible, no later file conversion can recreate it as observed geometry. Software may close, smooth or interpolate a region, but introduced geometry should not be confused with direct evidence.
Why the ASCAND camera stays still
Different methods observe different physical evidence
“3D scanner” is an umbrella term. It does not identify one universal mechanism.
Image- and silhouette-based methods
Camera-based methods analyze visual observations from changing viewpoints. Conventional photogrammetry commonly identifies corresponding image features and estimates their spatial relationship. Silhouette-based reconstruction uses the object’s outlines to constrain a volume consistent with the observed views.
These approaches depend differently on texture, separation from the background, viewpoint coverage and visibility. They should not be treated as interchangeable merely because both use images.
Projected-light and laser-triangulation methods
An active system projects a known pattern, point or line onto visible surfaces. The camera or sensor observes how that projection appears. With a known geometric relationship, triangulation can locate surface positions.
The result depends on calibration, alignment, visibility and how the surface reflects or transmits the projected light.
Time-of-flight and depth-sensing methods
Some systems estimate distance from the travel time or phase of emitted energy, or use integrated depth-sensing arrangements. Their working range, sampling, resolution and material response depend on the particular technology and implementation.
Contact measurement
A probe can record selected surface coordinates through physical contact. This is a different evidence process from optical capture and may be appropriate where discrete measured positions matter. It does not automatically provide complete surface appearance.
Volumetric imaging
Methods such as computed tomography can derive internal as well as external structure under appropriate conditions. They belong to the wider field of three-dimensional acquisition but involve equipment, physics and use cases very different from an optical turntable workflow.
ASCAND uses documented optical evidence: controlled camera observations, silhouette-informed Vision reconstruction and optional projected-line Laser reconstruction. The wider method families above provide context; they are not all ASCAND functions.
Understand ASCAND and photogrammetry
Explore laser triangulation
The method shapes what can be reconstructed
Different methods can produce different models of the same object because they do not collect identical evidence.
Six factors are especially important:
- Visibility: A surface must be observable from the capture relationship. Undersides, deep recesses and surfaces hidden behind other geometry may be missing.
- Sampling: The number and distribution of observations affect which changes in shape are represented.
- Surface response: Very reflective, transparent, translucent, dark or otherwise difficult surfaces can alter the recorded visual or projected-light evidence.
- Spatial relationship: Calibration, alignment, motion interpretation and coordinate consistency affect where observations are placed.
- Reconstruction assumptions: Algorithms use particular rules to turn evidence into geometry. Those rules may preserve, constrain, smooth, connect or fill different features.
- Intended use: The evidence sufficient for a classroom investigation or visual reference may be insufficient for a dimension-critical engineering decision.
It is useful to describe regions of a result more carefully:
- directly supported geometry is calculated from relevant observations;
- constrained geometry is limited by evidence such as silhouettes;
- interpolated or completed geometry is introduced to bridge gaps or create a continuous surface;
- missing geometry remains unsupported or absent.
These categories are conceptual, not automatic labels supplied by every file. They encourage a better question than “Does the model look right?”: Which parts of this result are supported by which observations?
Compare Vision, Laser and Combo
Review capabilities and limitations
A reconstruction, a representation and a file are not the same thing
The phrase “3D model” often hides several separate stages.
A reconstruction is the geometric result derived from observations and processing.
A point cloud stores sampled positions in three-dimensional space. Points may also carry attributes such as color or surface-normal information.
A mesh connects geometry into a surface, commonly with triangles or other polygons. Meshing can make a result easier to view, edit or fabricate, but the act of connecting points does not prove that every surface is accurate or observed.
Color and texture describe appearance. A model may resemble the object visually while still having incomplete or distorted geometry. Conversely, useful geometry may have limited appearance information.
A file format packages one or more representations. Choosing OBJ, STL, PLY, GLTF or another format does not itself improve the underlying evidence.
Post-processing adapts the representation to a task. Editing, cleaning, scaling, repairing, remodeling, adding tolerances or preparing for fabrication are downstream decisions—not proof that the original scan was complete.
Compare point clouds and meshes
Understand output formats and post-processing
Judge a scan against its intended use
There is no single meaningful answer to “Is this a good scan?” without knowing what the result is for.
Consider at least five separate dimensions:
Coverage
Are the regions needed for the task present? A model can look complete from one viewpoint while lacking an underside or recessed area.
Geometric fidelity
Does the reconstructed form correspond sufficiently to the physical object for the intended use? This question requires appropriate reference evidence. Visual plausibility alone is not dimensional verification.
Detail
Are relevant small features represented, or have sampling, visibility, smoothing or reconstruction choices removed them?
Surface appearance
Does the result include useful color or texture information? Appearance quality and geometric quality should be evaluated separately.
Downstream readiness
Can the chosen application use the representation and file? Does it require a watertight surface, verified scale, reduced complexity, editable CAD features, tolerances or additional repair?
A result may be strong in one dimension and weak in another. A scan suitable for viewing may not be suitable for replacement-part design. A useful geometric reference may still require direct measurement and CAD remodeling. A printable mesh is not automatically a safe or functional part.
Assess object suitability
Prepare a scan for 3D printing
ASCAND is one structured optical 3D scanning workflow
ASCAND applies the general observation-to-geometry model through a controlled turntable relationship.
The smartphone remains steady while the object rotates on the coded turntable. The video records changing views of the object and the visible orientation reference. This creates a structured sequence in which the system can relate observations to object rotation.
The documented scan types use different evidence:
- Vision uses object segmentation and silhouettes across orientations to constrain a visual hull through voxel-carving logic.
- Laser observes a projected line on visible surfaces and uses the known geometric relationship to support triangulation.
- Combo uses documented Vision and Laser evidence as complementary inputs.
- Multi-Scan and Multi Merge use separate captures in complementary physical orientations when another orientation exposes useful regions and sufficient overlap supports alignment.
The physical ASCAND system creates the controlled capture relationship. 3D-Scan.Online is the browser-based platform used to create scan projects, upload captured material, process it, review available results and download outputs. The hardware and platform have connected but distinct responsibilities.
ASCAND is therefore not accurately described as a generic image-to-3D generator or as conventional handheld photogrammetry. It is a structured reconstruction workflow whose results remain subject to visibility, object behavior, capture conditions, method limits and downstream requirements.
See the complete ASCAND workflow
Understand structured rotation and Gray Code
Continue with the question that matters to your project
Now that the shared definition is clear, choose the next concept by question:
- How is ASCAND different from conventional photogrammetry? Read ASCAND and Photogrammetry.
- Why does the smartphone remain fixed? Read Why the Camera Stays Still.
- Which method contributes which evidence? Compare Vision, Laser and Combo.
- Will a particular object work for my intended result? Use the Object Suitability assessment.
- What happens after reconstruction? Compare point clouds and meshes.
- What do I actually do? Move from conceptual knowledge to the complete workflow or a verified tutorial.
Understanding the evidence first makes method selection, result evaluation and downstream preparation more deliberate.