Technical
XGRIDS Lixel L3 White Paper: Validating 5 mm accuracy
Inside the XGRIDS Lixel L3 white paper: Explore the methodology, testing and results behind its 5 mm post-processed relative accuracy.
XGRIDS has published its Lixel L3 Point Cloud Accuracy Validation White Paper, detailing the methodology and data behind the flagship spatial scanner's 5 mm post-processed relative accuracy specification. Testing involved seven Lixel L3 units, five operators, and 20 independent scans in a challenging indoor environment, generating more than 1,500 point-pair comparisons. The headline result: 68% of absolute point-pair distance deviations were 4.2 mm or less under the documented conditions.
Key takeaways
Validated accuracy: XGRIDS' testing supports the Lixel L3's 5 mm @ 1σ (typical) post-processed relative accuracy specification.
4.2 mm P68: 68% of absolute point-pair distance deviations were 4.2 mm or less under the documented test conditions.
Robust methodology: Testing covered multiple L3 units, operators, and independent scans rather than relying on a single dataset.
Independent checks: A Leica TS60 established the reference measurements, with no GCP optimisation applied to the L3 point clouds.
Transparent evidence: The white paper provides professionals with the methodology and statistical context behind the headline specification.
Why it matters: Verified accuracy gives greater confidence when using L3 data for surveying, measurement, Scan-to-BIM and other accuracy-critical workflows.
Lixel L3: The Lixel L3 is XGRIDS' new flagship spatial scanner. Find out about the L3's key features and specifications in our in-depth guide.
Expert Insight
"For professionals evaluating reality capture technology, the headline accuracy figure is only part of the story. Understanding how that figure has been established - including the test environment, reference system, sample size and statistical methodology - provides much more useful context when assessing a scanner for a particular workflow. By publishing the methodology behind the L3's 5 mm relative accuracy specification, XGRIDS is giving users greater visibility of how its performance has been evaluated."
- Alex Williams, heliguy™ Technical Geospatial Specialist
Lixel L3 accuracy validation at a glance
Test detail | XGRIDS validation |
|---|---|
Accuracy specification | 5 mm @ 1σ (typical) post-processed relative accuracy |
P68 test result | 4.2 mm |
Scanners tested | 7 Lixel L3 units (both 16-channel and 32-channel) |
Operators | 5 |
Independent scans | 20 |
Test area | Approx 507 m² |
Reference equipment | Leica TS60 total station |
Point-pair comparisons | 1,524 |
Comparison distances | 5.40-44.22 m |
GCP optimisation | None |
Why validate mobile scanning accuracy?
Accuracy is an important consideration when selecting a spatial scanner for applications such as surveying, building digitisation, Scan-to-BIM, and engineering.
But the accuracy of a mobile SLAM scanner cannot be determined by looking at the LiDAR sensor specification alone.
The final point cloud can be influenced by sensor measurement, SLAM localisation and mapping, trajectory optimisation and environmental conditions. Loop closure and global optimisation can also adjust the scanner's estimated trajectory as the dataset is created.
XGRIDS therefore based its validation on the final post-processed point cloud output, assessing the performance of the complete scanning and processing system.
"Accuracy validation for a mobile laser scanning system should therefore take the final output point cloud as the object of evaluation, and systematically assess its relative accuracy and overall geometric consistency across varying spatial conditions and measurement distances."
- XGRIDS, Lixel L3 Point Cloud Accuracy Validation White Paper, 2026, p.3
How was the Lixel L3 tested?
XGRIDS tested seven Lixel L3 scanners - four 16-channel units with 120-metre range and three 32-channel units with 120-metre range - with five operators completing 20 independent scans.
The testing environment was as follows:
Item | Parameter |
|---|---|
Site type | Office - single floor |
Total area | Approximately 507 m² |
Layout | Two long parallel corridors, connecting corridor, multiple individual rooms, and a central stairwell. |
Corridor length | Up to 90 metres in length |
SLAM conditions | Frequent turns and loop closures |
The image below - Figure 1 from the report - is the top view of the test site point cloud and target distribution (the 13 targets included in the evaluation).
The characteristics of the test environment were particularly relevant to SLAM performance.
Long corridors can provide fewer lateral geometric constraints and increase the potential for accumulated longitudinal drift, while repeatedly moving between rooms creates numerous loop-closure decisions.
"In combination, these conditions are closer to the working reality of building survey and interior renovation projects than an open site or a short corridor would be."
- XGRIDS, Lixel L3 Point Cloud Accuracy Validation White Paper, 2026, p.6
Establishing an independent reference
To evaluate the resulting point clouds, XGRIDS positioned 13 circular high-reflectivity targets throughout the test site.
Their reference coordinates were measured using a Leica TS60 total station, with XGRIDS reporting a post-adjustment reference accuracy of approximately 1.5 mm.
The target layout created point-pair distances spanning approximately 5.4 m to 44.2 m, with more than half exceeding 20 m. This ensured the validation included longer measurements capable of exposing accumulated SLAM error.
Following each L3 scan, the targets were extracted from the point cloud and their centre coordinates calculated. Distances between the targets were then compared with the corresponding reference distances established using the total station.
Post-processed, but without GCP optimisation
An important distinction is that the validation relates to the Lixel L3's post-processed relative accuracy.
XGRIDS used its standard processing workflow, with Lixel Studio 4.1 listed as the processing software. However, no GCP optimisation was applied.
The targets therefore acted as check points for evaluating the resulting geometry - they were not used as control information to improve it.
This means the white paper evaluates processed L3 point-cloud data without subsequently optimising the geometry against the surveyed reference network.
More than 1,500 point-pair comparisons
Across the 20 scans, XGRIDS successfully extracted 257 of a possible 260 targets, representing a 98.8% extraction rate.
These generated 1,524 valid point-pair distance comparisons, ranging from 5.40 m to 44.22 m, with a median distance of 20.23 m. More than half of the comparisons were above 20 m.
For each target pair, the distance calculated within the L3 point cloud was compared with the independently established reference distance. The absolute difference between the two formed the deviation used in the accuracy analysis.
What were the Lixel L3 accuracy results?
XGRIDS pooled all 1,524 measurements from the seven scanners and 20 scans before calculating the results.
Percentile | Deviation |
|---|---|
P50 (median) | 3.0 mm |
P68 (≈1σ, primary metric) | 4.2 mm |
P85 | 6.4 mm |
P97 | 9.9 mm |
Mean | 3.5 mm |
The key result is P68 = 4.2 mm.
Put simply, 68% of the absolute point-pair distance deviations were 4.2 mm or less under the documented test conditions.
XGRIDS compares this result with the Lixel L3's specified 5 mm @ 1σ (typical) post-processed relative accuracy.
"The results show that, for the pooled sample, the absolute point-pair distance deviation at P68 is 4.2 mm, better than the specified 5 mm @ 1σ (typical)."
- XGRIDS, Lixel L3 Point Cloud Accuracy Validation White Paper, 2026, p.10
What does P68 mean?
P68 represents the 68th percentile. In this validation, it means that 68% of the measured absolute deviations were equal to or below 4.2 mm.
XGRIDS uses P68 as its primary metric and explains that, where signed errors are approximately normally distributed around zero, P68 is approximately equivalent to one standard deviation, or 1σ.
The result was also calculated on a pooled basis. All 1,524 measurements from every device and scan were combined into a single dataset before the percentiles were calculated.
This is an important consideration when comparing scanner specifications: similar headline accuracy figures may have been established using different environments, reference systems and statistical methods.
What does the validation mean for Lixel L3 users?
The significance of the white paper isn't simply the 4.2 mm result.
It provides transparency around how XGRIDS evaluated the L3's post-processed relative accuracy, including the number of scanners and operators, test environment, reference equipment, measurement distances, processing workflow and statistical methodology.
This gives surveying, AEC and reality capture professionals additional evidence and context when assessing the Lixel L3 for their own workflows.
There are also important limitations to consider.
XGRIDS states that the ≤5 mm @ 1σ specification represents expected performance under typical working conditions, while actual accuracy can be influenced by factors including site geometry, scan-path planning and environmental conditions.
The 4.2 mm P68 result should therefore be understood as performance under the specific test conditions documented in the white paper, rather than a guarantee that every dataset will produce the same result.
XGRIDS Lixel L3: Accuracy meets photorealism
The white paper provides the technical evidence behind the accuracy side of the Lixel L3 proposition.
XGRIDS' flagship spatial scanner combines 5 mm post-processed relative accuracy, integrated RTK and 4K HDR true-colour imaging, with a single capture capable of producing true-colour point cloud, mesh and LCC 3D Gaussian Splatting outputs.
This supports workflows ranging from precise measurement and modelling to photorealistic 3D representation across surveying and AEC, digital twins, and film and entertainment.
For professionals assessing the Lixel L3, the white paper provides valuable context behind the headline specification - showing not just the accuracy figure, but how that performance was evaluated.
