Quality review & project delivery | Drone Emotions
How to Read a DJI Terra Quality Report
Turn the DJI Terra quality report into a practical review of your reconstruction. Understand the main indicators, compare control and check points, inspect 2D and 3D outputs, and prepare a separate quality check for LiDAR point clouds.
Read the DJI Terra quality report in three passes
A useful review moves from the project context to the diagnostic indicators and finally to the actual files you will deliver.
1. Confirm the project
Check the source dataset, sensor, software version, processing run and coordinate reference. Make sure the report belongs to the same reconstruction as the outputs you are reviewing.
2. Read the evidence
Review image alignment, control and check-point information, relevant warnings and processing settings. Record questions that need investigation rather than judging the project from one attractive number.
3. Inspect the deliverable
Open the exported map, surface, model or point cloud in its receiving application. Check the required coverage and measurements against the project’s agreed acceptance criteria.
Define acceptance before processing. Write down the required area, output type, coordinate system, horizontal and vertical tolerances, and independent checks. This makes the report useful for a decision instead of a collection of numbers without a target.
Where to find the report and which document to use
Use the completed mission’s report controls and result-directory access in your installed Terra version. DJI’s output-directory guide identifies report data with aerotriangulation results and report directories for 2D and 3D reconstruction. The exact directory layout and interface can vary with the workflow and release.
Keep the report with the corresponding outputs. If you processed the same dataset several times, record the run date and changed settings so that the files cannot be confused.
DJI’s report reference PDF
The locally hosted PDF is a useful reference for report terminology across aerotriangulation, 2D reconstruction, 3D reconstruction and LiDAR processing. DJI’s download center dates its quality-report PDF to October 10, 2023.
Use it as a dated reference; field names and available sections may differ in a later release.
Current software documentation
Use DJI’s online manual and the instructions for your installed release when looking for a control, output directory or processing option. Our Download page brings the main setup resources together.
DJI references: reconstruction results and directories · current download center.
DJI Terra report indicators: a quick interpretation
Read the units, reference and processing stage for each number. A value in pixels answers a different question from a difference measured in meters.
| Indicator | What it describes | What to check next |
|---|---|---|
| Calibrated images | Images successfully reconstructed during aerotriangulation. | Locate any missing coverage. |
| Connected components | Separate image networks in the solution. | Check gaps between adjacent survey areas. |
| Reprojection error RMS | Image-space residual, normally expressed in pixels. | Keep it separate from ground-coordinate error. |
| Georeferencing RMSE | Agreement of solved camera positions with recorded POS. | Identify the reference behind the statistic. |
| GCP residuals | Fit at points used to constrain the adjustment. | Review markings and coordinate inputs. |
| Checkpoint errors | Differences at points excluded from adjustment. | Inspect individual errors and their distribution. |
| GSD / output resolution | Ground pixel spacing or selected output sampling. | Check detail separately from positional accuracy. |
| Warnings and settings | Messages and recorded processing choices. | Document their relevance to the delivery. |
On a small screen, swipe across the table to view every column.
This table summarizes commonly encountered terms. Use the definition shown in your report and the applicable DJI documentation when a field differs.
Investigate patterns as well as averages. Ask whether a concern is isolated, concentrated at an edge, associated with one flight or spread across the dataset. Record the affected area before choosing a corrective action.
Definitions: DJI report reference and DJI photogrammetry accuracy checks.
GCP residuals and checkpoint errors: assess them separately
Ground control points influence the adjustment. Check points are reserved for evaluating the result without that adjustment role. A close fit to control does not replace an independent check.
Review the control inputs
Confirm point identifiers, coordinate units, axis order, horizontal reference and height reference. Inspect the image markings at points with unusual residuals.
Record the survey method and quality of the reference coordinates, because a comparison inherits uncertainty from the reference data.
Review the independent checks
Identify which points were excluded from optimization. Inspect both summary statistics and individual differences, and document where the points sample the delivery area.
Do not repeatedly remove inconvenient points or change their role simply to obtain a more attractive summary.
DJI’s photogrammetry accuracy guidance distinguishes reprojection error from georeferencing agreement and evaluates control and checkpoint statistics separately. Follow the documented workflow for marking and optimization in the version you use.
For survey planning, coordinate references and RTK context, continue with our Accuracy, GCP & RTK Guide. This page focuses on reviewing the evidence after processing.
Understand RMSE with a simple worked example
Root mean square error summarizes a set of differences by squaring each difference, averaging those squares and taking the square root. It uses the units of the differences. Large differences have more influence than small ones.
Illustrative elevation differences
These are invented values for explanation, not a DJI benchmark or a measured Drone Emotions project.
Suppose four independent checks have elevation differences of +2 cm, −2 cm, +4 cm and −4 cm.
Mean signed difference = 0 cm
RMSE = √((2² + (−2)² + 4² + (−4)²) / 4) = √10 ≈ 3.16 cm
The signed mean is zero because positive and negative differences cancel. The RMSE still shows the magnitude of the differences.
For a real review, retain the point count, individual differences and their locations. Check which axis or combined quantity the reported RMSE represents. A vertical statistic alone does not describe horizontal performance.
Compare the result with the agreed project requirement and the quality of the reference measurements. A single summary value does not describe every location in the delivered area.
Inspect 2D maps and 3D models alongside the report
Review the actual exported files at a useful scale. A completed reconstruction and its report should lead to a deliverable check in the application where the data will be used.
2D orthomosaic and surface review
- Confirm that the required boundary is covered and that gaps are documented.
- Inspect seams and areas around buildings, vegetation, water and moving objects.
- Check raster resolution, no-data areas, coordinate reference and units.
- Review surface elevations at suitable independent locations.
- Verify that the receiving GIS reads the files and georeferencing correctly.
3D model and point-cloud review
- Inspect required surfaces, edges and areas hidden from the cameras.
- Look for holes, duplicated surfaces or geometry that needs explanation.
- Check scale, orientation, coordinate reference and output format.
- Test the measurements or comparisons relevant to the intended use.
- Open the delivery in the client’s viewer or processing workflow.
Resolution and accuracy answer different questions. A fine sampling interval can show detailed imagery while a coordinate bias remains. Include geometric checks as well as a visual inspection.
When reading the report’s timing and processing settings, use them to explain how the run was produced. They are helpful for repeatability and workstation planning; faster completion is not an acceptance criterion for positional quality.
Further reading: DJI reconstruction outputs · USGS imagery calibration guidance. For hardware planning, use our System Requirements Guide.
Read a DJI Terra LiDAR quality report in its own context
A LiDAR point-cloud review follows the acquisition, positioning, processing and elevation-check evidence relevant to that mission. Do not transfer an image reprojection criterion to a LiDAR point cloud.
Acquisition and processing
Identify the payload, flights, positioning information and processing choices recorded for the run. Read warnings and note which parts of the site or acquisition they affect.
Reference and height checks
Confirm the horizontal and height references for the point cloud and surveyed points. Review point roles and the elevation differences at independent checks.
Surface and classification
Inspect overlaps between flights, representative profiles and the surfaces needed for the delivery. If ground classification or an elevation model is required, review those outputs separately.
DJI’s support article describing accuracy control and check from Terra V3.9.0 explains an elevation-focused workflow. Control points adjust elevations; check points assess the point cloud at their locations. The article recommends flat, open, unobstructed surroundings for these points.
Check the current manual for your release and payload before applying that procedure. Elevation control should not be interpreted as proof that every horizontal or surface requirement has been verified.
Record checkpoint count, individual differences and their spatial distribution alongside any overall statistic. Inspect the actual cloud or derived elevation model at the places relevant to the project.
DJI reference: LiDAR accuracy control and check. For sensor planning, see our Supported Drones & Sensors Guide.
Troubleshooting: turn a report concern into a test
The following are investigation prompts, not automatic diagnoses. Preserve the original run and record one change at a time so that a comparison remains meaningful.
| Observed concern | First investigation | Next action |
|---|---|---|
| Missing reconstructed coverage | Check which input images and survey areas are involved. Inspect acquisition coverage and source quality. | Keep a map of the missing area before deciding whether to reprocess or reacquire. |
| Unusual pixel residuals | Review the affected image observations and point markings against the source photographs. | Change only a documented input or setting, then compare a separate run. |
| Low control errors but poor checks | Review point roles, reference coordinates and the locations of independent checks. | Assess the independent evidence rather than optimizing the control summary alone. |
| A consistent elevation difference | Review height definitions, transformations, units and the reference survey. | Investigate the cause before applying an offset to a deliverable. |
| A good summary but visible defects | Inspect the exported geometry and the areas where the required data is incomplete. | Record the defect and its effect on the intended use. |
| Differences after an update | Compare the exact software versions, source files, settings and output types. | Retain both runs and consult the relevant release notes. |
On a small screen, swipe across the table to view every column.
For a processing failure, record the exact error and stage and consult DJI’s reconstruction error guidance. For a version comparison, use our Versions & Release Notes Guide.
A practical DJI Terra delivery checklist
Use this as an internal review record and adapt it to the project specification. It does not replace the assessment method required by a client or applicable surveying standard.
- Match the files. Confirm that source data, processing run, report and exports belong together.
- Record the environment. Note the Terra release, sensor, processing date and relevant settings.
- Confirm references. Record coordinate system, units, height reference and any transformations.
- Review coverage. Compare reconstructed and exported coverage with the required area.
- Read warnings. Document unresolved messages and the affected data or processing stage.
- Review point roles. Retain the control/check distinction and the reference survey information.
- Inspect individual results. Record questionable points and spatial patterns as well as summary numbers.
- Open the outputs. Test the required files in the intended receiving application.
- Check the intended measurements. Use the independent verification agreed for the deliverable.
- Explain limitations. Identify missing areas, excluded data or unresolved quality concerns.
- Record the decision. Document acceptance, reprocessing or further acquisition with a reason.
- Archive the evidence. Keep the report, run settings and accepted deliverables together.
Prepare a useful support request
Include the Terra version, sensor, intended deliverable, exact warning or concern, processing settings and a short description of what you expected. Identify the relevant report section and output area. For a comparison, explain what changed between runs.
DJI Terra quality report FAQ
Where is the DJI Terra quality report?
Start from the completed mission and its result-directory access in your installed release. DJI’s output-directory reference identifies report data alongside reconstruction results. Check the current manual if your interface or folder layout differs.
Is the report PDF the same as my project report?
No. The reference PDF explains report terminology. Your project report contains information from a specific processing run; keep it with the corresponding dataset and deliverables.
Does a completed reconstruction mean the project is accepted?
Completion shows that the processing stage finished. Acceptance also requires the coverage, files, references and quality checks agreed for the intended use.
Can I use a pixel error as a ground accuracy value?
Keep the reported units and reference clear. An image-space diagnostic should not be presented as a measured ground-coordinate accuracy.
Is there one RMSE value that is good for every project?
Set the requirement for the intended measurement and deliverable. Record the reference quality, sample count and spatial coverage before comparing a statistic with that requirement.
What if I have no independent reference measurements?
Describe the evidence that is available and the checks that remain unverified. Do not label an output independently validated when that comparison was not performed.
Should I remove a point that worsens the report?
Investigate the point first. Correct a demonstrated input or marking error and document the change. Preserve justified checks even when they reveal a problem with the result.
Can I compare reports from different processing runs?
Yes, provided the dataset, settings, point roles, references and output types are recorded. Identify the differences so that a changed statistic can be interpreted in context.
Should I review a LiDAR report like a photogrammetry report?
Use the report and documentation for the actual mission type. Prepare the relevant positioning, elevation and surface checks for the LiDAR delivery.
Do I need the same quality checks for visualization and surveying?
Choose checks around the intended use. A visual presentation and a measurement deliverable have different acceptance needs; document which requirements your review addresses.
Official references and related DJI Terra guides
This page provides original explanations, an illustrative calculation and suggested review checklists. The official references below define software terms and procedures; their applicability depends on your release and workflow.
Need help planning your Terra workflow?
Tell Drone Emotions which sensor, outputs and deployment you use. Include the software version and the report question when requesting setup or workflow advice.
Original educational guide by Drone Emotions, an authorized reseller. Reviewed October 5, 2026. DJI supplies the application and its documentation. Check the current documentation for changes in report fields and procedures.