9.4 Aerial-Ground Fusion Processing

What This Mode Produces

Aerial-Ground Map Fusion merges a ground scan with drone aerial imagery into one model that covers interior and ground-level geometry, the roof, and the surrounding site. The ground scan supplies precise interior detail. The drone supplies the roof and the areas no handheld scanner can reach. Use it for campus facilities, construction sites, and heritage or landmark documentation where roof and aerial context are required. Skip it for interior-only work, at any scale.

Plan the Site and the Flight

The ground scan and the drone flight are joined by shared points on the ground called aerial-ground fusion points. Placing them well is the entire job. Plan them before anyone powers on a device.

• Survey the site first. Use satellite imagery or an aerial photo to lay out the site before you arrive.
• Choose the fusion points: at least 4, covering 8 approach directions. Pick open, unobstructed areas spread evenly across the site to serve as the drone's takeoff and landing points. These same spots become the fusion points that tie air to ground. On most mid-size jobs, 4 to 5 points is right. Keep adjacent points no more than about 165 ft apart, and on dense sites plan roughly one point per quarter acre. Points clustered at one end of a large site do not constrain the far end, and the model will drift there.
• Pick feature-rich spots, not blank pavement. The fusion matches geometry visible in both datasets. A paved area with planters, signage, bollards, or building corners gives it far more to lock onto than an empty square.
• Confirm each point is visible from both air and ground. The drone and the scanner must both be able to see the same fixed targets, roughly 65 to 130 ft from the point, since that shared view is what the bridge photos will capture.
• Keep the point areas unchanged between visits. If the scan and the flight happen on different days, parked vehicles or staged materials that move between them degrade the fusion at that point.

Capture the Ground Scan

Scan the site as usual, with two additions that create the fusion points.

1. Mark a fusion point at each takeoff and landing location. At each chosen spot, mark an aerial-ground fusion point and circle it once so the scanner captures a complete cloud around it. LCC Studio identifies these points automatically during processing, so they do not need names.
2. Route the scan path through every fusion point. Plan the walking path so it passes through each point, and close two loops within 6.5 to 10 ft of each point. The double loop closure at the point is what gives the fusion a reliable tie.
3. Cover the rest of the site normally, and watch the live cloud. Follow standard scan technique for the interior and ground-level exterior between the fusion points. Monitor the real-time point cloud as you go: drift, misalignment, layering, or large-scale noise in the live view means that stretch needs a rescan before you leave the site.

Fly the Drone Mission

The mission is the overhead mapping flight. The goal is complete, high-overlap coverage of every surface you want in the model.

The core principle is overlap: every ground object of interest needs more than 85% overlap across the image set. Higher overlap produces a more complete reconstruction. For large areas with complex structures, add 1 to 2 passes at higher altitudes; the height difference between adjacent passes must not exceed twice the lower altitude.

Capture the Vertical Bridge Photos

The vertical bridge is the step that decides whether the ground scan and the drone mission actually merge. The mission photographs the site from above; the ground scan captures it from eye level; on their own, the software has no way to connect the two. The bridge fills that gap. At each marked point, the drone climbs a vertical column, photographing the same scene continuously from just above the ground all the way past the mission's flight height. That unbroken run of images is what lets the fusion lock the two datasets together.

Capture a bridge column at every marked takeoff and landing point:

1. Start low, just above the ground. Position the drone over the marked spot 1.5 to 3 ft off the ground, and aim the camera at a fixed target 65 to 130 ft away that is visible to both the drone and the scanner, in a texture-rich direction. Every photo in the set aims at that same target.
2. Climb straight up past flight height, shooting as you go. Enable timed capture at 0.5-second intervals and rise slowly and steadily from just above the ground to above the mission's flight altitude, collecting 30 to 40 photos across the full climb and holding at least 85% overlap between one shot and the next. The lowest photo sits within 3 ft of the ground; the highest sits above the mission flight height. Control the climb speed by hand so the interval shots keep their overlap.
3. Fly the column from 3 directions. Repeat the climb aimed from the left, the center, and the right of the target, so each of the 3 directions is its own 30-to-40-photo set at 85% overlap. On drones with Point of Interest and Orbit functions, set the target as the POI and let Orbit hold the aim while you manage the climb.

Details that matter on the bridge:

• How high to rise between shots. Space the 30 to 40 photos evenly across the climb so consecutive shots overlap by at least 85%. On a tall column this is a small step: for a climb to 250 ft, a photo roughly every 6 to 8 ft; on a lower flight the step is shorter.
• How many photos per point. About 30 to 40 per direction, across 3 directions, so roughly 90 to 120 photos at each takeoff and landing point. Budget the flight time and the battery for that.
• The descent counts too. Photos taken during a continuous ascent or a continuous descent are all valid, so a climb-and-descend pass at one point can cover two directions.
• Same camera rules as the mission. Wide-angle mode, distortion correction disabled, RTK Fixed held for every shot, and no blurred, overexposed, or underexposed frames. Photos without a Fixed RTK record are thrown out in processing.
• Do not mark or name control points for the bridge photos. These are imagery, not survey points. You do not add a control point or a name for them. Just fly the column and shoot.
• Keep the bridge photos with the mission photos on L series and K2. On those scanners, every bridge photo goes into the same folder as the drone mission photos; there is no separate upload for them. On PortalCam, the takeoff and landing photo sets are uploaded as their own per-point folders alongside the aerial photos, so keep each point's set in its own folder from the start. See Section 7.

Aerial Data Requirements

Before processing begins, LCC Studio rejects an aerial image set that does not meet these. On L series and K2 jobs the set is the mission photos and the bridge photos together, in one folder. On PortalCam jobs the takeoff and landing photo folders are uploaded per fusion point alongside the aerial photos; the counts and formats below apply to the full combined set.

Pairing the drone RTK with the ground scan

The bridge photos tie the two datasets together visually. Positioning ties them together in space, and how you supply that positioning depends on the ground device. A switch in the reconstruction parameters, on by default, controls whether the drone's own RTK data is used at all. Turning it off makes the system ignore the drone's RTK completely.

Match the pairing to the equipment you actually have before the flight, because the wrong pairing is not something processing can repair.

Before committing a multi-hour run, run the XGRIDS quality inspection tool against the assembled data. It verifies RTK status, coordinate systems, point distribution, photo counts, and capture duration in minutes, not hours. The full tool workflow and the nine checks it supports are documented in 6.4 Aerial-Ground Fusion.

Run the Fusion

Processing takes two inputs: the scanner project file (or raw segments), and the drone photos.

1. Organize the inputs first. Keep the scanner data in one place. On L series and K2 jobs, put every drone photo, mission and bridge, into a single image folder; do not sort the bridge photos into separate subfolders. On PortalCam jobs, keep the aerial photos in their folder and each fusion point's takeoff and landing photos in their own per-point folders; uploading only the aerial photos without the marked fusion points and their photo folders degrades the fusion or fails the reconstruction.
2. Create the project. Click Create and select Aerial-Ground Map Fusion. This mode is distinct from Map Fusion and Single Model and cannot be switched later without reprocessing.
3. Add the ground data. Under Ground Data, add the raw scanner data: up to 10 raw segments, planned at 20 to 30 minutes each, under a 200-minute total. Map Fusion output is not accepted. LCC reads the fusion point markers from the scan and identifies them automatically. The interface shows the device type it detects.
4. Add the aerial data. Under Aerial Data, select the drone photos: on L series and K2, the single folder holding the full set; on PortalCam, the aerial photos plus the takeoff and landing photo folders for the corresponding fusion points.
5. Set parameters and start. Quality Standard for a first run. Maximum Gaussian Points applies per block in this mode and the system auto-adjusts it. Set the drone RTK option to match what the drone actually recorded, per the pairing table in Section 6: on when the flight carried RTK, off on the PortalCam and non-RTK-drone path. Start the reconstruction.

Check the fusion at the ground-to-air transition

When processing finishes, open the model and check the one place a bad fusion shows up: where the ground data meets the aerial data. Find a wall or a tall feature that both the scanner and the drone captured, then follow it from the ground, which came from the scanner, up to the roofline, which came from the drone. It should read as one continuous surface.

A bad fusion at that spot looks like a horizontal step, a sideways offset, a doubled or ghosted wall, or a visible break at the height where the two datasets meet. Any of those means the bridge column at that point did not give the fusion enough overlap between ground and air. That point needs its bridge re-flown; the break cannot be fixed in post.

Processing requirements