Section 9: Aerial and Ground Fusion

Feb 9

Combining drone imagery with ground-level scan data to produce complete 3D models of sites that cannot be fully captured from either perspective alone.

Quick Reference

Pre-Flight Requirements

Requirement	Detail
RTK	Enabled on both drone and scanner throughout collection
Coordinate system	WGS84 on both drone and scanner
Drone images	JPG/JPEG only, minimum 1024 x 768, consistent resolution
Image count	100 to 10,000 per project
Main flight overlap	85% side overlap and 85% forward overlap
GSD	1 cm or below
Gimbal angle	45 degrees
Fusion point count	4 to 5 open areas, evenly distributed across site
Photos per fusion point	30 to 40 images, from 3 directions, ground level to flight height

L2 Pro Drone Mode — Flight Parameters

Parameter	Recommended	Maximum
Altitude	30 m	50 m above tallest structure
Speed	3 m/s	5 m/s
Flight distance	Unlimited	Unlimited
Compatible drones	DJI Matrice 300 RTK, DJI Matrice 350 RTK	—
Aerial accuracy	6 cm	—
Power-on sequence	Drone first, then L2 Pro	—

Which Approach to Use

Approach	Device	Output	Processing
Aerial-Ground Map Fusion	PortalCam, K1, or L2 Pro + mapping drone	3D Gaussian Splatting model	LCC Studio
L2 Pro Drone Mode	L2 Pro mounted on Matrice 300 or 350	LiDAR point cloud	LixelStudio

12.1 — When to Use Aerial-Ground Fusion

Ground-level scanning produces precise, detailed geometry of every space a scanner can physically access. Drone imagery captures rooftops, upper building facades, and large open terrain that no amount of ground coverage can document. Aerial-ground fusion combines both into a single coherent model, and it is the right workflow any time either perspective alone would leave a meaningful gap in the deliverable.

The most common use cases are: campus and institutional facilities where buildings are surrounded by grounds; construction sites where exterior progress, grading, and roof structure require documentation alongside interior scan data; heritage sites where full architectural envelopes and fine surface detail must coexist in one output; and any project where building rooftops, mechanical equipment, or structures above accessible height are in scope. If a specification requires a complete exterior envelope, aerial-ground fusion is not optional.

The XGRIDS ecosystem supports two technically distinct approaches. Choosing the right one depends on your hardware configuration and the required deliverable type.

Aerial-Ground Map Fusion (LCC Studio) uses a mapping drone to capture photogrammetric imagery, which LCC Studio fuses with ground scan data to produce a 3D Gaussian Splatting model. This is the right choice for photorealistic visualization, client presentations, and web delivery. It works with the PortalCam, K1, and L2 Pro on the ground side, and any RTK-capable mapping drone on the aerial side.

L2 Pro Drone Mode (LixelStudio) mounts the L2 Pro directly on a DJI Matrice 300 RTK or Matrice 350 RTK and collects LiDAR data from the air. Processing happens in LixelStudio and produces a georeferenced point cloud. This is the right choice when the deliverable requires measured aerial geometry for BIM, CAD, or infrastructure documentation at survey-grade precision.

These two approaches can be run in parallel on the same project when both a point cloud and a 3DGS model are required. Section 1 of this guide covers the broader question of when to use each pipeline — reviewing that section before planning a combined aerial-ground project will help you organize your field data collection correctly from the start, since the required data differs between pipelines.

Troubleshooting 12.1

Problem: Uncertain which approach to use for a given project. Determine the deliverable requirement first. If the client or specification asks for measurements, CAD, or BIM integration, the answer is L2 Pro Drone Mode with LixelStudio. If the deliverable is a visual 3D model for review, presentation, or web sharing, the answer is Aerial-Ground Map Fusion with LCC Studio. If both are required, plan to run both workflows concurrently during the same site visit, since ground scan data collected with RTK active is compatible with both pipelines.

12.2 — Planning an Aerial-Ground Project

Aerial-ground fusion requires more pre-site coordination than a standard ground-only scan. The alignment between aerial and ground data is established at the time of field collection. Problems that could be corrected on site with an additional ten minutes of work become unrecoverable once you are back at the office.

Selecting Fusion Point Locations

The most important planning decision is where to place your fusion points. These are the drone takeoff and landing locations that serve as the spatial bridges between aerial and ground data. XGRIDS recommends selecting 4 to 5 open areas evenly distributed across the site.

Each location must be open and unobstructed so the drone can safely execute a vertical takeoff and landing sequence. It must also be visually distinctive — both from the air and from ground level — so the fusion algorithm has identifiable features to work with. Featureless surfaces such as uniform grass, empty asphalt, or smooth concrete should be avoided as fusion locations. Spots with visible texture, surface material changes, permanent fixtures, or painted markings are reliable choices. Sketching the planned fusion point locations on a site map or aerial photograph before departure is a practical habit that prevents gaps during collection.

Plan your ground scanning route so that it passes through each fusion point location. This is a firm requirement, not a preference.

Coordinate System Planning

Both the ground scanner and the drone must use WGS84. Set this before any data collection begins, in both LixelGO (or LCC Scan for PortalCam) and on the drone's flight controller. A mismatch in coordinate systems is one of the most common causes of fusion failure and cannot be corrected in post-processing.

Hardware Checks

Before departure, confirm that all software is current. Ensure the LCC+ point cloud option is selected in LixelGO before scanning with a K1 or L2 Pro for the LCC Studio workflow. Drone images must be saved as JPG or JPEG — RAW format is not supported by LCC Studio. If the drone saves both RAW and JPEG simultaneously, configure it to JPEG-only before the mission to avoid unnecessary data and conversion steps. If using a multifocal drone lens, all images for a single processing project must use the same focal length throughout.

Troubleshooting 12.2

Problem: Suitable fusion point locations are difficult to find on site. Featureless sites such as flat agricultural land or uniform paved areas are the most challenging. In these cases, place temporary physical markers — survey targets, painted stakes, or distinct objects — at planned fusion locations before scanning begins. These create the shared visual features that both the scanner and drone camera need to establish registration. Remove or note the markers in your project documentation for client reference.

Problem: The scanner route cannot pass through all planned fusion points due to physical obstacles. Redesign the fusion point locations before scanning begins rather than improvising in the field. Any fusion point that the scanner route cannot access will fail during processing. If the site layout makes it genuinely impossible to connect all planned fusion points with a single walking route, the project may require multiple separate ground scan sessions that are merged using Map Fusion before the aerial-ground fusion step.

12.3 — Ground Data Collection for Fusion

Ground scanning for fusion projects follows standard technique with two specific additions: deliberate coverage at fusion point locations, and RTK active throughout the session.

K1 and L2 Pro

In LixelGO, set the coordinate system to WGS84 and confirm RTK is enabled before starting the scan. Follow standard LCC collection procedures throughout. When your route arrives at a fusion point location, slow down and complete 2 loop closures within 2 meters of the takeoff and landing point. This creates sufficient scan density and trajectory overlap at the exact location where the drone will need to register its aerial data. A single pass is not enough. After completing the two loops, continue along your planned route to the next fusion point.

PortalCam

The PortalCam ground scan workflow follows the standard PortalCam procedure. At each drone takeoff and landing location, mark an Aerial-Ground Map Fusion point in the LCC Scan app. After placing and marking each fusion point, walk a full circle around it before continuing. This 360-degree pass around the fusion point provides the scan density and coverage that the fusion algorithm needs at that location.

Fusion point placement rules: choose locations with good lighting, clear texture, and unobstructed open areas. PortalCam orientation at the fusion point must match the drone camera direction. The allowed tolerance for returning to a previously marked position across sessions is less than 0.5 m distance, plus or minus 20 degrees of orientation, and approximately 10 cm of height difference.

Across All Devices

Scan the remainder of the site using normal technique. RTK must remain active throughout any portion of the ground scan that will be submitted for aerial-ground fusion processing. Areas without RTK coverage during scanning cannot be reliably fused with aerial data.

Troubleshooting 12.3

Problem: RTK signal drops or is unavailable in parts of the site. Segment the scan so that the portion without RTK is handled as a separate segment that does not include any fusion points. Only the segments that cover fusion points and the areas directly connecting them require RTK for the fusion to work. If RTK is unavailable at a fusion point location specifically, that location cannot be used as a fusion point — move it to an area with confirmed RTK coverage.

Problem: PortalCam fusion point was not marked during the scan session. If you reached a fusion point location but forgot to mark it in the LCC Scan app, you cannot retroactively add it. Complete the scan session, then plan a second PortalCam session that specifically covers the missed fusion point. The second session can be merged using Map Fusion before the aerial-ground fusion step, provided it overlaps with the original scan by at least 15 meters in a feature-rich area.

Problem: After completing the 2 loop closures (K1 / L2 Pro), the trajectory shows drift in that area. Drift in a loop closure area indicates insufficient environmental features for the scanner to track against — likely an open area with few distinct surfaces. This is a site characteristic, not a technique error. To compensate, scan in tighter loops and vary your height by crouching or extending the scanner upward during the loops. You can also place temporary physical markers in the open area to create trackable features.

12.4 — Aerial Data Collection (LCC Studio Workflow)

This section covers aerial data collection for the LCC Studio Aerial-Ground Map Fusion pipeline. The drone used for this workflow is a standard RTK mapping drone with a photogrammetric camera. The L2 Pro Drone Mode workflow is covered separately in Section 12.5.

Drone Configuration

Both the Matrice 300 RTK and Matrice 350 RTK are compatible with the Aerial-Ground Map Fusion workflow in LCC Studio when paired with an appropriate photogrammetric camera payload. The Matrice 400 will add to this lineup when released. Power on the drone and set the coordinate system to WGS84 on the flight controller before beginning the mission.

Main Flight Mission

Use a smart grid flight pattern over the survey area, framed on the controller map to cover the full site. Set capture to oblique mode with the gimbal angle at 45 degrees. Set GSD to 1 cm or below — higher resolution produces more accurate model reconstruction. Set flight height above the tallest structure in the survey area to ensure safety throughout the mission. In advanced flight settings, set both side overlap and forward overlap to 85 percent. This level of overlap is the minimum required for reliable reconstruction. Higher overlap improves model completeness.

Tap Start and allow the drone to fly the planned route automatically. Monitor the drone continuously throughout the mission and pause to make corrections if needed.

Fusion Point Photo Sequences

After completing the main flight mission, visit each of the 4 to 5 fusion point locations on the ground. At each location, do the following:

Point the drone camera at fixed objects that are visible to both the drone and the ground scanner — permanent fixtures, building corners, surface markings, or other identifiable features.
Capture 30 to 40 photos at that location, spanning from ground level up to the full flight height. The lowest photos should be taken at approximately the same height as the ground scanner's collection height. The highest photos should match the planned flight altitude.
Maintain 85 percent overlap between adjacent photos throughout the sequence.
Repeat from 3 different directions at each fusion point. Shooting from a single direction is not sufficient for consistent fusion success.

These vertical photo sequences are the spatial bridge between the aerial mission and the ground scan data. Without them, the registration algorithm has no shared reference between the two data sets and fusion quality degrades significantly or fails outright.

Troubleshooting 12.4

Problem: Main flight mission completes but you run out of battery before completing the fusion point photo sequences. The fusion point sequences are more critical to registration quality than the main flight. If battery is limited, prioritize completing at least one 3-direction sequence at each fusion point before finishing the main mission nadir pass. Always plan battery reserves specifically for the fusion point sequences — do not treat them as optional coverage to add at the end if time allows.

Problem: GSD of 1 cm requires a very low flight altitude, but obstacles on site make that unsafe. Fly at the lowest safe altitude achievable given site constraints and accept the resulting GSD. Document the actual GSD in your project notes. A higher GSD will produce a less detailed model but will not cause the fusion to fail. The 85% overlap requirement is more critical to reconstruction success than achieving exactly 1 cm GSD.

Problem: Wind conditions make it difficult to maintain stable hover for the fusion point sequences. Complete the fusion point sequences in the most sheltered locations possible given the site. Increase the number of images captured per sequence — capturing 50 or more instead of the recommended 30 to 40 reduces the impact of any individual blurred or poorly framed image. If wind conditions are severe, reschedule the aerial collection for a calmer time rather than accepting degraded data.

12.5 — L2 Pro Drone Mode

L2 Pro Drone Mode mounts the L2 Pro scanner on a DJI Matrice 300 RTK or Matrice 350 RTK and collects LiDAR point cloud data from the air. This is a fundamentally different workflow from the photogrammetry-based LCC Studio approach — the output is measured geometry rather than a visual reconstruction.

The specified aerial accuracy for L2 Pro Drone Mode is 6 cm, compared to 3 cm for handheld operation. The difference reflects the additional positioning uncertainty introduced by altitude and drone motion. For most AEC documentation and site survey work, 6 cm aerial accuracy is fully adequate. Ground control points can be added during LixelStudio processing to constrain the aerial data to real-world coordinates if tighter absolute positioning is required.

Hardware Overview

The L2 Pro Drone Mounting Kit contains three components: the mounting bracket, the gimbal adapter ring, and a custom Type-C cable. The L2 Pro's standard RTK module is removed before mounting — the drone provides RTK positioning through its own onboard system, selected as Aerial RTK type in LixelGO.

The RTK antenna that the L2 Pro system uses for drone mode mounts on top of the M300 or M350. This is a separate antenna from the standard L2 Pro RTK module and connects through the mounting system.

Mounting the L2 Pro

Attach the bracket backplate to the L2 Pro body using M2.5 x 5 screws.
Connect the custom Type-C cable to the gimbal adapter using M2.5 x 8 screws.
Install the gimbal adapter ring on the M300 or M350 gimbal port.
Mount the bracket to the drone using the existing gimbal holes with M3 x 8 screws. Confirm the directional arrow on the bracket points toward the gimbal adapter ring before tightening.
Verify the Type-C cable connection between the L2 Pro and the gimbal adapter is secure. The drone powers the L2 Pro through this connection.

LixelGO Configuration

Open LixelGO and configure the startup screen before beginning the scan:

Setting	Value
Mount type	Drone
RTK type	Aerial
GNSS mode	DJI
UAV ellipsoid	WGS84 (or CGCS2000 — must match the drone controller)

LixelGO must be used to start the scan. The project file will not record the mount type, RTK type, and coordinate system correctly if the scan is started any other way. LixelStudio reads these values from the project file during processing and cannot accept manual corrections after capture.

Power-On Sequence

Power on the drone first, then the L2 Pro. The L2 Pro draws power from the drone through the mounting connection.

Flight Execution

Set the flight route on the DJI remote controller and begin the mission at a constant speed. The L2 Pro scans continuously throughout the flight.

Parameter	Recommended	Maximum
Altitude	30 m	50 m
Speed	3 m/s	5 m/s
Flight distance	Unlimited	—

When the drone completes the mission, stop the scan in LixelGO before powering off the L2 Pro. Power off the L2 Pro first, then the drone.

Data Organization

Before processing, organize project files into the following folder structure. This structure is required for LixelStudio to correctly associate the aerial and ground data.

ProjectName/
├── Drone/               # L2 Pro drone scan project file
│   └── project_data/
└── Lixel/               # Ground scan project file(s)
    └── project_data/

Place the project folder from the drone scan in the Drone subfolder and the ground scan project folder in the Lixel subfolder. If there are multiple ground scan sessions, each gets its own subfolder within Lixel.

The XGRIDS Extract Preview Poses Tool

XGRIDS provides four supplemental files used to prepare data for aerial-ground fusion processing in the older K1 / L2 Pro workflow. After organizing the Drone and Lixel folders, copy these four files into the project directory and launch the Extract Preview Poses application. This tool processes the drone image metadata to generate the camera pose information that LCC Studio needs for fusion. Once it completes, use the included visualization tool to confirm that all camera poses were correctly identified before submitting the project for processing.

If you do not have these files, contact XGRIDS support directly. They are provided as part of the workflow support package for L2 Pro Drone Mode.

Troubleshooting 12.5

Problem: LixelGO does not show the Drone option for mount type. LixelGO version 1.2.0 or above is required for drone mode. Update the app if the option is not present.

Problem: The drone powers on but the L2 Pro does not initialize. Confirm the Type-C cable between the L2 Pro and the gimbal adapter is fully seated on both ends. If the cable connection is secure and the L2 Pro still does not initialize, power off both devices, reconnect the cable, and repeat the power-on sequence (drone first, then L2 Pro).

Problem: LixelStudio processes the drone scan as a handheld scan. The mount type was not correctly recorded in the project file, most likely because the scan was started outside of LixelGO. There is no way to correct this after capture. The session must be rescanned with LixelGO active and mount type set to Drone.

Problem: The directional arrow on the mounting bracket was not aligned toward the gimbal adapter ring before tightening. The bracket will need to be remounted. Do not fly with the bracket incorrectly oriented — the sensor orientation data recorded by the L2 Pro during flight is calculated based on the expected mounting direction, and misalignment will corrupt the point cloud geometry.

12.6 — Processing Aerial-Ground Fusion in LCC Studio

This section covers the LCC Studio 3DGS processing pipeline using the Aerial-Ground Map Fusion project type. LCC Studio Premium is required.

LCC Studio supports up to 10 segments of ground scan data in a single Aerial-Ground Map Fusion project.

Project Type Selection

Open LCC Studio and select Generate Model. At the top of the dialog, select Aerial-Ground Map Fusion. This is a separate option from standard Map Fusion. Do not use the Map Fusion project type for aerial-ground projects — the data upload structure is different and the project will not process correctly.

Uploading Ground Data

Under Ground Data, add the scan files from your Lixel device. For ground scans collected with a device from the Lixel L2 family (which includes a built-in panoramic camera), also upload the corresponding panoramic video file. For K1 and L2 Pro, the panoramic video option is grayed out — proceed directly without it. Add each ground scan segment separately if the project used multiple sessions.

LCC Studio will automatically detect how many Aerial-Ground Map Fusion points are associated with the ground data.

Uploading Aerial Data

Import order matters. Follow this sequence:

Import all photos captured during the main drone flight first — the full grid mission imagery goes in as a single set.
Then import the photos taken at each takeoff and landing site in sequence, one fusion point at a time.

LCC Studio uses the import order to associate the fusion point photo sequences with the correct ground-side fusion point markers. Importing in the wrong order will produce incorrect associations and degrade or prevent successful fusion.

For PortalCam workflows, the expected folder structure groups takeoff/landing images by fusion point within the Drone folder:

Drone/
├── DJI_Link_1/          # takeoff/landing images at Fusion Point 1
│   ├── image1.jpg
│   └── ...
├── DJI_Link_2/          # takeoff/landing images at Fusion Point 2
│   ├── image1.jpg
│   └── ...
└── [main grid images]   # all remaining aerial mission images

For K1 and L2 Pro workflows, the Drone folder contains all drone images, with the XGRIDS Extract Preview Poses tool having been run beforehand to generate the required pose files.

Scene Parameters

Configure scene parameters as you would for any standard 3DGS project. For large aerial-ground projects, Standard or Slow reconstruction quality is recommended over Fast, since the added complexity of two data sources increases the risk of alignment artifacts in a faster reconstruction. Enable Portability if the output will be shared via the web viewer. PPR should be set to Low for outdoor aerial-ground projects to reduce sky artifacts and improve edge clarity on structures.

Ground Control Points

If the deliverable requires georeferenced absolute accuracy, add GCPs through the LCC Studio GCP workflow during project setup. The Aerial-Ground Map Fusion interface shows a dedicated Takeoff Point / GCP section when this project type is selected. GCPs constrain the model to real-world coordinates and generate accuracy reports suitable for professional deliverable packages.

Processing Time

Aerial-ground fusion is the most computationally demanding workflow in LCC Studio. Expect 24 to 48 hours or more for a full-site project on recommended hardware. Smaller projects with limited aerial imagery can complete more quickly. Do not schedule fusion processing for times when the workstation is needed for other tasks.

Troubleshooting 12.6

Problem: "Aerial-Ground Map Fusion" option is not visible in LCC Studio. This feature requires LCC Studio Premium. Verify your subscription level. If Premium is active and the option still does not appear, confirm you are running a current version of LCC Studio and check the XGRIDS support documentation for version-specific feature availability.

Problem: LCC Studio reports that no fusion points were detected after importing ground data. The scanner did not correctly record the fusion point markers during data collection. For PortalCam, this means the Add Fusion Point step was skipped in the LCC Scan app during scanning. For K1 and L2 Pro, this typically means the LCC+ point cloud option was not selected in LixelGO before scanning. Neither condition can be corrected in post-processing. The ground data collection must be repeated.

Problem: Photos imported in the wrong order at the fusion point step. Delete all aerial data from the project and reimport following the correct sequence: all main mission images first, then each fusion point takeoff/landing set one at a time in the order they correspond to the ground-side fusion point markers. There is no way to reorder images within the project after import.

Problem: Insufficient Memory error during processing. Aerial-ground fusion has significantly higher RAM and VRAM requirements than single-scene 3DGS. Enable Low-Memory Mode in LCC Studio and reduce the Maximum Gaussian Splats slider. Close all other applications during processing. If the error persists, the workstation may not meet the hardware requirements for projects of this scale. Splitting the project into a smaller area with fewer ground segments and fewer aerial images will reduce the memory load.

Problem: Processing completes but the ground and aerial portions are visibly offset. This is a registration failure. The most common causes are mismatched coordinate systems between the scanner and drone, weak RTK during one or both collection sessions, or insufficient overlap and coverage at fusion points. Review your field notes to confirm coordinate system alignment, then check the RTK quality indicators in the project data. If RTK was weak during the ground scan at any fusion point location, that fusion point cannot provide a reliable anchor and its registration quality will be poor.

12.7 — Processing in LixelStudio (Point Cloud Merge)

When aerial data was collected using L2 Pro Drone Mode, all processing happens in LixelStudio. The output is a merged point cloud combining aerial and ground LiDAR data. LixelStudio version 3.3 or higher is required.

Processing Individual Sessions

Process the drone scan and each ground scan session separately first. In LixelStudio, click Project Processing and select the project file. For drone session files, LixelStudio automatically reads the mount type (Drone) and ellipsoid from the project file and applies the correct parameters. You do not manually specify that the data came from a drone — this is why correctly recording the mount type in LixelGO during collection is essential.

Repeat for each ground scan session.

Merging the Point Clouds

Once all individual sessions are processed, use LixelStudio's Map Fusion feature to merge the aerial and ground point clouds into a single registered dataset. The same Map Fusion principles apply here as for merging multiple ground sessions: shared RTK data and overlapping scan regions provide the alignment anchor, and each segment is registered to a common coordinate frame. Section 9 of this guide covers the LixelStudio Map Fusion workflow in detail.

Evaluating the Merged Result

After merging, inspect the combined cloud at boundary zones where aerial and ground data intersect. A clean merge shows a seamless transition between dense ground-level geometry and the aerial LiDAR capture above. Visible double surfaces, offset geometry, or density gaps in the overlap region indicate that the two data sets did not align correctly.

Troubleshooting 12.7

Problem: LixelStudio does not detect the drone mount type from the project file. The mount type was not recorded during data collection. See the troubleshooting note in Section 12.5 — this requires recollection of the drone session.

Problem: The merged point cloud shows double surfaces or geometric offset at the aerial-to-ground boundary. Review RTK quality for both sessions. If RTK was active and strong during both sessions and the coordinate systems matched, the problem may be insufficient scan density at the boundary — the area where the drone LiDAR data transitions to ground LiDAR data needs overlap to register correctly. Plan drone flight paths so that the lowest altitude pass covers the same physical area that the ground scanner captured near building facades and at outdoor transition zones.

Problem: Aerial point cloud is significantly noisier than the ground point cloud. Some additional noise in aerial LiDAR data is expected and reflects the 6 cm vs. 3 cm accuracy specification difference between drone and handheld modes. For deliverables where the aerial portion needs to be cleaner, apply noise filtering in LixelStudio after merging. If the noise is extreme rather than marginal, review the flight parameters — speeds above the recommended 3 m/s or altitudes above 30 m increase noise in the aerial data.

12.8 — Evaluating Aerial-Ground Results

A well-executed aerial-ground fusion produces a model where the transition between ground-level and aerial data is invisible. Building facades blend from ground-scanned detail at street level to aerially captured geometry at upper floors without seams or offsets. Rooftop geometry is clean and matches the footprint visible in the ground-level scan. Texture and color are consistent across the full elevation of a structure when the 3DGS pipeline is used.

Common Failure Modes

Symptom	Likely Cause	Prevention
Reconstruction failure at project start	Fusion point markers not recorded during ground scan	Mark fusion points at every drone takeoff location during scanning; do not skip the loop closures or circle step
Ground and aerial data visibly offset	Mismatched coordinate systems; weak RTK during one session	Confirm WGS84 on both devices before collection; verify RTK lock before starting
Fusion point photo sequences not connecting to ground data	Photos taken from only one direction, or fewer than 30 per fusion point	Capture 30 to 40 photos from 3 different directions at each fusion point
Blurry or low-detail aerial geometry	GSD too high; flight altitude too high; insufficient overlap	Fly at lowest safe altitude; target GSD of 1 cm or below; confirm 85% side and forward overlap
Aerial portion present but ground is absent or distorted	RTK not active during ground scan	RTK must be enabled throughout all ground scan segments intended for fusion
Insufficient Memory error	Workstation below hardware requirements for this scale	Enable Low-Memory Mode; reduce Maximum Gaussian Splats; upgrade RAM if recurring
L2 Pro drone scan treated as handheld in LixelStudio	Scan not started through LixelGO	Always use LixelGO with mount type set to Drone and RTK type set to Aerial

A Note on First-Time Projects

Aerial-ground fusion has more variables and more potential failure points than any other XGRIDS workflow. Running a small test project on a low-stakes site before committing this workflow to a critical deliverable is worth the investment. Even an experienced scanner will benefit from working through the full collection-to-processing pipeline on a familiar site before deploying on a complex or time-sensitive project.

Troubleshooting 12.8

Problem: The final model looks correct overall, but there are gaps or missing geometry on specific roof surfaces. The main flight mission did not capture those surfaces with sufficient overlap or at a sufficient angle. Oblique flight at 45 degrees captures facades and sloped surfaces well, but flat rooftops with raised parapets or equipment may require a second nadir pass directly overhead. Plan a supplemental nadir pass for projects where rooftop documentation is specifically required.

Problem: The 3DGS model looks good in LCC Studio but loses quality after export or web publishing. Check the Portability setting. If Portability is off, the exported model may be too large or complex for standard web delivery. Enable Portability for web-shared outputs. For export quality, review the export settings — 3D Tiles and other formats have resolution and geometric error parameters that control output fidelity. Section 11 of this guide covers advanced export configuration.

Problem: Client finds it difficult to interpret the aerial-ground model because interior and exterior do not visually connect. This is a presentation issue rather than a technical one. Use LCC Studio to create a cropped exterior-only view and a separate interior view, then use the Portal feature in LCC Studio to link them with jump points. Section 14 covers the Portal workflow. This approach allows clients to navigate from an aerial overview directly into specific interior spaces, which substantially improves deliverable usability for large-site projects.

Sources: Lixel L2 Pro User Manual v2.3; LCC CyberColor Studio User Manual v1.10.0; LCC Scanning Guide v9.0; XGRIDS Products FAQ; XGRIDS LCC 1.9.0 Release Highlights; XGRIDS Aerial-Ground Map Fusion Tutorial transcripts (K1/L2 Pro and PortalCam workflows); XGRIDS Aerial-Ground Fusion Workflow transcript (E38 Survey Solutions)

Ian Jack