Turn an inspected capture into a Vision reconstruction

Turn an inspected capture into a Vision reconstruction

An ASCAND Vision Scan uses a recorded sequence of the rotating object without requiring the optional laser workflow. 3D-Scan.Online interprets the observations in that source and produces a result for inspection.

Begin only with a video that passed the pre-upload recording check. The object, coded reference, framing, focus, background and lighting should already have been evaluated across the intended capture. Processing can interpret recorded evidence; it cannot recreate views that were cropped, obscured, interrupted or never recorded.

This tutorial describes the verified workflow boundary from source selection to result review. Follow the current platform documentation for the exact interface controls, supported files, access requirements and processing choices. The linked Knowledge article explains the reconstruction principle without turning internal implementation details into customer instructions.

Check the source before processing

Understand Vision reconstruction

Select the inspected original recording

Choose the original saved recording that you inspected immediately after capture. Avoid using a visually similar copy when you cannot confirm whether it has been shortened, recompressed, reframed, stabilized, filtered or otherwise changed.

Before continuing, confirm:

1. The file opens and plays from beginning to end.
2. It contains the intended uninterrupted capture interval.
3. The camera viewpoint remains steady.
4. The object remains stable relative to the turntable.
5. The complete object remains inside the frame.
6. The required coded reference remains observable.
7. Object boundaries remain distinguishable from the background.
8. Focus and illumination remain suitable across the rotation.

Also confirm that Vision is an appropriate evidence path for the intended result. Vision constrains an outer form from recorded observations. Deep recesses, internal voids or geometry that never changes a visible outline may remain unresolved. A clean source does not remove that geometric boundary.

Use the current documentation to verify supported file type, maximum file size, transfer method and any restrictions on edited video. Those requirements can change independently of the reconstruction principle.

If the source does not pass, correct the capture cause and rerecord. Uploading a known defect consumes time without adding evidence.

Return to video inspection

Confirm the current Vision workflow

Open the current 3D-Scan.Online workflow and follow its documented sign-in and project procedure. Before submitting anything, verify that the project is intended for a Vision Scan and that every required configuration choice corresponds to the physical capture.

Depending on the current interface, this may include a turntable configuration, scan type, processing choice or project information. Use the names and options shown in current documentation. Do not select a familiar-looking default merely because it allowed a previous project to proceed.

The physical configuration matters because the platform must interpret observations within the applicable capture relationship. The scan method matters because Vision, Laser and Combo use different evidence. A project label is therefore not merely administrative metadata.

At this checkpoint, answer four questions:

• Is this the correct source video?
• Is Vision the intended method?
• Does the selected physical configuration match the capture?
• Are the current access and submission requirements satisfied?

If any answer is uncertain, pause before upload or submission. Consult the current platform documentation. Do not infer a current interface step from an older screenshot.

Upload the source and verify the submission

Use the verified controls in the current platform. The bounded sequence is:

1. Choose the inspected original source.
2. Begin the documented transfer.
3. Confirm that the transfer reaches its verified completed state.
4. Review the project, method and configuration selections.
5. Submit or start processing once using the documented control.
6. Retain the original video until the result has been reviewed.

Upload completion and processing submission are different events. A file may have reached the platform before a reconstruction job has been accepted or started. Use the current interface state rather than assuming that a progress indicator means the entire workflow is complete.

Avoid repeated submission when the status is merely unclear. Duplicate actions can create ambiguity about which project or result belongs to the inspected source. Follow the current recovery guidance or use support if the interface does not establish what happened.

Do not close, refresh or leave the page based on generic web behaviour. Whether a transfer must remain open and whether managed processing continues independently are operational facts that must be verified for the current platform.

How managed processing works

Let the platform interpret the observations

At a public conceptual level, Vision processing can be understood as a sequence:

1. Decode the source into usable image observations.
2. Interpret rotation evidence so observations can be associated with object orientation.
3. Separate object and background evidence in selected views.
4. Use those observations to reject unsupported parts of a candidate volume.
5. Extract an exterior representation from the remaining volume.
6. Prepare a result that can be previewed, evaluated or processed further.

This sequence explains why capture quality matters, but it is not a list of interface controls. Exact implementation, parameters, intermediate representations and available processing branches depend on the verified current system.

Vision does not directly see every surface as a measured depth sample. In silhouette-informed voxel carving, a view can show that a region must be empty because it belongs to the background. Combining such constraints from multiple orientations leaves the largest volume consistent with the observations—the visual hull.

That makes the method useful for reconstructing an outer form, but it also creates a structural limit: geometry that remains hidden behind the visible exterior cannot be carved away merely because processing runs longer or at a different display resolution.

Explore the silhouette-to-volume sequence

Interpret status without guessing

The current interface may distinguish several operational states. Whatever labels are used, keep these meanings separate:

• Transfer incomplete: The full source has not reached the platform.
• Submitted or queued: The source has been accepted for managed work, but a final result is not yet available.
• Processing: The system is evaluating the source or building an intermediate or final representation.
• Completed: The documented processing path has produced a result or next action.
• Failed, stopped or uncertain: The workflow needs the current error guidance, a diagnostic step or support.

A completed upload does not mean reconstruction is complete. A completed reconstruction does not mean the geometry is suitable. A failed state does not by itself identify whether the cause is the source, configuration, platform, processing path or account state.

Do not estimate completion from file size, object complexity or an earlier project. Do not repeatedly resubmit solely because the job takes longer than expected. Use only current, published timing and recovery guidance.

If the interface reports a problem, preserve the source, project identity, visible status and any documented diagnostic information. Avoid publishing private account information when requesting help.

Diagnose Vision Scan problems

Review the geometry—not only the completion message

Open the available result using the current review controls. Inspect the model from multiple directions rather than judging it from the initial view.

Check:

1. Project identity: Is this result associated with the intended source and settings?
2. Subject identity: Does the result represent the object you recorded rather than background or support geometry?
3. Orientation and scale plausibility: Is the model positioned and sized as expected within the documented coordinate model?
4. Overall form: Does the outer envelope agree with the object across front, side, rear, top and oblique views?
5. Silhouette-supported regions: Are the major boundaries that were visible in the recording represented coherently?
6. Missing or excess geometry: Are parts absent, enlarged, bridged or filled where the observations may have been weak or ambiguous?
7. Surface artifacts: Are there spikes, rough regions, steps, isolated components or unstable boundaries?
8. Representation state: Are you reviewing a preview, point cloud, mesh or other output, and what further work does that representation require?

Do not confuse a smooth preview with verified geometry. Display shading, point size, interpolation, texture or viewer settings can change appearance without changing the underlying reconstruction. Similarly, a closed-looking surface is not evidence that hidden concavities were recovered or that dimensions are certified.

If a defect corresponds to cropping, movement, focus, lighting, background separation or coded-reference visibility, return to capture diagnosis. If the source appears sound but the result is incomplete or structurally unsuitable, use the Vision troubleshooting route and reconsider whether Vision observed the geometry the task requires.

Understand Vision strengths and limits

Decide what the result supports

A completed Vision Scan is now evidence you can evaluate. Choose the next action from the result—not from the completion message.

• The result supports the intended next step: Continue to the detailed review, representation, export and download procedure.
• The source contained a correctable capture problem: Correct the station, repeat the full-rotation check, record a new source and create a new Vision result.
• Processing completed but the result shows an unexplained symptom: Follow the Vision diagnostic path before changing multiple variables.
• The required geometry was not observable through the Vision evidence: Compare Laser, Combo or additional-orientation workflows against that specific limitation.

Another method is not automatically better. Laser observations introduce their own visibility, material and setup conditions. Combo requires evidence-aware reconciliation. Multi-Scan can add coverage only where another orientation observes new geometry and still provides adequate overlap for registration.

The correct result is not necessarily the most detailed-looking preview. It is the representation whose known evidence, limitations and remaining preparation are appropriate for the intended use.

Review, export and download the model

Diagnose a Vision result
Compare Vision, Laser and ComboAn ASCAND scan video preserves a sequence of views while the object rotates in front of a steady camera. The recording stage should capture the physical relationship you already verified: the same camera position, object placement, turntable configuration, framing, background and lighting.