Turn captured observations into usable 3D geometry
When you upload an ASCAND recording, the video becomes the source material for a Scan Project in 3D-Scan.Online. The platform does more than change one file format into another. It interprets a controlled sequence of observations, reconstructs the geometry supported by those observations and prepares representations that can be reviewed and used downstream.
This work continues as a managed background process after upload. The available results depend on the selected processing path, the evidence present in the recording and the stages that complete successfully. Processing can refine captured information, but it cannot fully recover surfaces that were never observed.
Follow the processing sequence
Reconstruction is not ordinary file conversion
A conventional converter reads an existing model and writes the same underlying content in another format. ASCAND processing begins earlier: the uploaded source is a video, not a finished 3D object.
The system first turns the recording into usable observations. Frames are extracted and prepared, relevant object regions are identified, and the applicable reconstruction methods derive geometric evidence. That evidence can become point-based geometry, be refined or reconciled with complementary sources, and later support polygonal surface generation and export.
Each representation is therefore the result of another interpretation or processing stage:
Video → frames → normalized observations → geometric evidence → point cloud → mesh → export model
Changing the final format cannot add a viewpoint, reveal a hidden underside or repair an unstable capture. The quality of every later representation remains constrained by the evidence that entered the pipeline.
How 3D scanning reconstructs geometry
Processing in six stages
1. Create the Scan Project and upload the source
Begin with the checked recording from the capture stage. In the current 3D-Scan.Online workflow, create or open the appropriate Scan Project, select the applicable processing path and upload the source material.
The exact controls, accepted files and access conditions must match the live platform. Keep the original recording until the complete workflow has finished. It remains the primary acquisition evidence and may be useful if a different processing choice or a corrected project setup is required.
2. Validate the source before reconstruction
The uploaded material must be usable as an ordered observation sequence. File integrity is only one part of that requirement. The object must remain observable, the rotation must provide the intended coverage and method-specific evidence—such as the coded region or projected laser profile where required—must be present.
Automated quality control may identify certain input or setup problems, but it cannot certify a successful result in advance. A recording can be technically readable while still lacking the observations needed for reliable geometry.
3. Prepare the observations
Processing extracts frames from the source and prepares them for geometric analysis. This can include image normalization, identifying the relevant region, separating the object from its surroundings and interpreting rotational or calibration references used by the selected path.
This stage creates more consistent inputs for later reconstruction. It does not erase the effects of blur, changing reflections, severe exposure problems or movement between the object and its support. Preparation improves the interpretability of recorded observations; it does not replace good acquisition.
4. Reconstruct the selected evidence
ASCAND supports distinct reconstruction paths. A Vision workflow derives geometry from image-based observations, including visual relationships, boundaries and silhouettes. A Laser workflow detects the projected laser profile and uses the calibrated camera–laser relationship to reconstruct illuminated surface points. A Combo workflow can use complementary evidence from more than one source where the documented configuration supports it.
These paths are not merely different export settings. They interpret different physical evidence and can produce different intermediate geometry. The system preserves that distinction so evidence can be evaluated before it is reconciled.
5. Refine geometry and generate representations
Reconstruction produces point-based geometric evidence. Point-cloud processing can evaluate neighborhoods, remove unsupported outliers, normalize density and prepare the geometry for later use. Where complementary reconstructions are available, fusion can evaluate agreement and disagreement rather than treating every point as equally reliable.
Mesh generation is a separate surface-reconstruction stage. It constructs a polygonal surface from suitable point-cloud input and can apply cleanup or optimization. A mesh is useful for many downstream workflows, but it is not automatically more truthful than the points from which it was built.
6. Return the results for review
Completed representations are returned to the project so they can be previewed, assessed and downloaded where the current workflow provides them. A completed job means that a processing path finished; it does not mean that every part of the object was captured perfectly or that every possible output exists.
Review the geometry for coverage, shape, noise, holes and surface artifacts before deciding what to use next. A weak region may point to an acquisition limitation, a method-specific problem or a need for a complementary object orientation.
The selected path determines which evidence is processed
Vision
Vision processing works from image-derived observations. It can use visual relationships, object boundaries, silhouettes and other image information to infer geometry. It benefits from consistent images, useful surface information and complete viewpoint coverage. Featureless, reflective or transparent regions can reduce the reliability of image-based interpretation.
Laser
Laser processing requires the optional, correctly configured Laser Extension workflow. It identifies the projected profile across the object and reconstructs surface points through calibrated triangulation. This provides a distinct source of geometric evidence, but only where the laser is visible and detected reliably. Alignment, calibration, illumination, occlusion and surface behavior remain important.
Combo
Combo processing uses complementary evidence rather than assuming that one method is universally correct. Laser-derived, vision-derived or silhouette-supported geometry can contribute different information. Where they agree, confidence can increase; where they disagree, the system must evaluate the conflict. Combo cannot recover a surface that is absent from every contributing capture.
Vision, Laser and Combo compared
Point clouds and meshes are different stages
A point cloud represents reconstructed geometry as sampled three-dimensional points. It preserves direct or inferred observations without requiring them to form a continuous skin immediately. This makes point clouds valuable for inspecting coverage, density, noise and the relationship between different evidence sources.
Point-cloud processing can remove isolated outliers, evaluate local consistency, normalize density and prepare independent reconstructions for fusion. Its role is refinement. It cannot turn an unobserved surface into measured geometry.
A mesh is a polygonal surface constructed from suitable point-cloud input. Mesh generation estimates how neighboring geometric samples should form a continuous surface, then may clean or optimize that surface for later use. The result can be easier to view, print or import into other tools, but meshing introduces a surface interpretation. A closed-looking mesh is not proof that every region is supported equally well.
Choose the representation for the task. Point clouds retain sampled geometry for inspection and further processing; meshes provide connected surfaces for workflows that require them.
Understand point clouds and meshes
Processing continues as a managed background job
3D reconstruction is computational work that does not need to run entirely in the browser or on the capturing smartphone. After upload, 3D-Scan.Online coordinates the applicable processing stages and returns completed results to the project.
Processing duration can vary with the source material, selected evidence path, workload and the representations being generated. Do not infer a guaranteed duration from a previous scan or an older interface. Follow the current project status information provided by the live platform.
Because processing is asynchronous, avoid treating a closed browser window, an unchanged preview or a single status observation as proof that a job has failed. Use only the current documented status and support workflow.
Learn about platform processing and delivery
Processing can refine evidence, not replace it
Filtering, confidence evaluation and complementary reconstruction can make the pipeline more robust. They still operate on captured observations.
Processing can:
- interpret consistent video frames as geometric evidence;
- derive independent Vision or Laser representations where applicable;
- reduce unsupported outliers and improve point-cloud consistency;
- reconcile complementary evidence when compatible reconstructions exist;
- construct a surface mesh from suitable point-based geometry.
Processing cannot:
- reveal a surface hidden from every capture;
- restore detail destroyed by severe blur or poor focus;
- make unstable acquisition geometrically consistent;
- guarantee reliable laser geometry from a missing or incorrectly aligned profile;
- turn transparency or reflections into dependable surface evidence automatically;
- make every object suitable for every reconstruction path.
When a result is incomplete, identify whether the limitation began in capture, calibration, method selection, reconstruction or surface generation. The most useful correction may be a better source recording or complementary orientation—not simply another export.
Before you start processing
- Review the complete source recording, not only its first frame.
- Confirm that the object stayed stable, framed and in focus.
- Select the processing path that matches the captured evidence.
- For Laser or Combo work, verify the current hardware, calibration, alignment and safety instructions.
- Allow the upload to complete using the current platform workflow.
- Check the current access terms without relying on legacy credit information.
- Preserve the original capture until all required results have been reviewed.
This checklist reduces avoidable processing attempts. It does not guarantee a complete or artifact-free model.
When processing completes, review the result
The project may contain one or more reviewable representations, depending on the selected evidence path and the stages that completed successfully. Inspect coverage, shape and artifacts before choosing a download or planning complementary work.