A sequence of photographs can become a 3D model of a facade, wall or construction element. The key is to capture from different positions, keep common areas between images and use clearly identified references.

From capture to model

Terrestrial photogrammetry reconstructs an object's shape from photographs taken at ground level. The software finds shared features between images to calculate camera positions and the geometry of the scene.

The workflow we are developing connects ROLAVICam for capture, AutoTags for identifying references and ROLAVIGeoStudio for preparing data for Agisoft Metashape. The goal is to preserve the connection between photographs, markers and coordinates throughout the project.

Photogrammetry in action

Watch the ROLAVICam demonstration that accompanies this workflow.

ROLAVICamSound off · Auto repeat
Watch on YouTube ↗

Video by ROLAVI GNSS. Playback is muted on this page.

Schematic: from capture to photogrammetric adjustment

This schematic summarizes the relationship between Take Point / Mark, 90% overlap, GNSS, sensors and AutoTags. Solid arrows show the data brought together in GeoStudio; the dashed branch shows visual-inertial support when compatible VIO recordings exist.

ROLAVI · Photogrammetry: capture, references and preparationDownload figure (SVG) ↗
Photogrammetry: capture, references and preparationConceptual workflow: surveyed AutoTags, capture data and GNSS go to GeoStudio. IMU and camera support is conditional. Agisoft processes images and GCPs; independent points verify accuracy.A · FIELD | ROLAVICam1Take Point / MarkSurvey the AutoTag center. Keep its IDand coordinates; import an externalcontrol point when needed.Surveyed AutoTags → GCPs2Configure overlap: 90%Capture sharp photographs from differentpositions. Keep the ID, capture time andtag observations.Photographs + time + tags3GNSS / RTK positionRecord coordinates, time and quality. IfRTK weakens, visual-inertial support canmaintain relative continuity.Global reference + quality4IMU + camera (VIO)*Combines sensors and vision to estimaterelative motion. Requires actual VIOtracking and recording.Support when RTK is weakCapture ID + time + referenceB · OFFICE | ROLAVIGeoStudio1Match the recordsPhoto + time + GNSS + AutoTags; VIO when available.2Geotag with a valid referenceAssociate camera positions only with verified timing and mounting geometry.3Prepare the GCPsID + surveyed coordinates + marker-center observations in the images.C · RECONSTRUCTION | Agisoft Metashape ProAlign images → adjust cameras and GCPs → generate the 3D model.VERIFY THE RESULTIndependent check points + known distances.
Photogrammetry: capture, references and preparationConceptual workflow: surveyed AutoTags, capture data and GNSS go to GeoStudio. IMU and camera support is conditional. Agisoft processes images and GCPs; independent points verify accuracy.A · FIELD | ROLAVICam1Take Point / MarkSurvey the AutoTag center. Keep itsID and coordinates; import anexternal control point whenneeded.Surveyed AutoTags → GCPs2Configure overlap: 90%Capture sharp photographs fromdifferent positions. Keep the ID,capture time and tag observations.Photographs + time + tags3GNSS / RTK positionRecord coordinates, time andquality. If RTK weakens,visual-inertial support can maintainrelative continuity.Global reference + quality4IMU + camera (VIO)*Combines sensors and vision toestimate relative motion. Requiresactual VIO tracking and recording.Support when RTK is weakB · OFFICE | ROLAVIGeoStudio1Match the recordsPhoto + time + GNSS +AutoTags; VIO when available.2Geotag with a validreferenceAssociate camera positions onlywith verified timing andmounting geometry.3Prepare the GCPsID + surveyed coordinates +marker-center observations inthe images.C · RECONSTRUCTION | AgisoftMetashape ProAlign images → adjust camerasand GCPs → generate the 3Dmodel.VERIFY THE RESULTIndependent check points + knowndistances.
Data and referencesConditional VIO support*
Figure 1. Conceptual terrestrial photogrammetry workflow. AutoTags connect images; their surveyed coordinates allow them to serve as GCPs. GeoStudio prepares references for adjustment in Agisoft.

*The dashed branch applies only if the version and capture include recorded VIO. The IMU alone does not replace RTK. This diagram does not imply the entire integration is available in every version or guarantee a particular accuracy.

In ROLAVICam: survey the AutoTag centers with Take Point / Mark and retain their identifiers. Set overlap to 90% for capture and check actual photographic coverage.

When RTK quality drops: camera and IMU can work together as VIO to estimate relative motion. The IMU alone accumulates drift; visual tracking, time records and references are needed to link that trajectory to the project coordinate system.

In ROLAVIGeoStudio: combine photographs, times, GNSS and AutoTag observations. Geotagging requires valid, synchronized camera positions. Surveyed tags provide GCPs for later adjustment in Agisoft, while other points are reserved for accuracy checks.

What do AutoTags do?

ROLAVI AutoTags are 55 coded markers for A4 printing. Their identifiers help recognize the same reference in different photographs. A printed marker does not know its coordinates: its role depends on the data you assign to it.

Visual reference

Without measured coordinates, it helps identify the same point across images.

Control point

With measured coordinates, it can help reference and adjust the model.

Check point

Its coordinates are held out of the adjustment to evaluate the result.

Distribute markers across the working area and at different heights where possible. Avoid clustering them in one corner or placing all of them along a single line.

Fieldwork in 4 steps

  1. Prepare the markers. Download the PDF, print at 100% and keep the complete pattern. Fix each AutoTag to a flat, stable surface. Use a different ID for each point.
  2. Measure the references. If you need a georeferenced model, survey the tag centers with GNSS or a total station. Retain each ID, coordinate system, units and height reference. The Mark → AutoTags workflow lets you organize records and import CSV controls in compatible versions.
  3. Capture overlapping photographs. In ROLAVICam, set the overlap target to 90% and check that neighboring photos share enough detail. Move slowly and change position around the object: rotating from a single spot does not provide the same geometry. Keep images sharp and avoid changing the zoom within a sequence.
  4. Review before leaving. Check that tags are complete and legible in several images from different angles. Retake blurred photographs and cover hidden areas. A configured percentage alone does not guarantee actual overlap.

Practical example: if the printed target is 07, keep the identifier Tag_07 in the records and processing. In the Surveying workflow, tags01 is the classification code; it does not replace each marker's unique ID.

Imported coordinates must already refer to the marker center. Do not subtract pole or prism heights again. Do not mix ellipsoidal and orthometric heights without the appropriate conversion.

From ROLAVIGeoStudio to Agisoft

Keep the original data together: photographs, capture records and the AutoTags folder or files. In ROLAVIGeoStudio, use the Cellphone workflow supported by your version to prepare images and controls.

Check that every ID matches its tag, detections are correctly centered and the coordinate system is correct. Then prepare the project for Agisoft Metashape Professional, where photographs are aligned, the block is adjusted and the model is generated.

ROLAVI AutoTags use their own pattern. Use the detections or exports from the ROLAVI workflow; do not assume Agisoft's coded target detector will recognize these markers directly.

GNSS provides global references; photographs provide the geometry. The antenna coordinate is not automatically the camera center: using camera positions must account for the assembly's geometry and synchronization.

How do I know the model is reliable?

Reserve measured points for independent checking and examine their horizontal and vertical differences. Also check a known distance and look for gaps, duplicated surfaces or distorted areas.

If GNSS quality drops, properly surveyed control points can provide a reference for the later adjustment. Tags without coordinates help connect images, but do not by themselves turn a capture without signal into a georeferenced model with centimeter accuracy.

A good result combines sharp photographs, adequate coverage, correct controls and verification. That is the foundation of the terrestrial workflow we are building with ROLAVI.

You can also follow the terrestrial videogrammetry workflow, in which each frame is linked to its time within the video.

Google · Camera and IMU · Further reading: Agisoft · Image capture · Control and check points · Targets and scale bars.