3.1 Startup and Movement

Startup

The startup position is the reference frame for the entire session. A bad start means every later loop closure works against a bad baseline, and there is no way to repair it in processing.

Before You Set the Device Down

• Check the LiDAR window and every camera lens. Clean, dry, and fog-free. Devices fog when carried between hot and cold environments, and a fogged lens looks clean at a glance.
• Confirm battery at three bars or better. A device that force-stops mid-scan leaves a partial segment and no way to resume, because no XGRIDS scanner can resume from a breakpoint.

Initialization

1. Set the device on flat, stable ground. The L2 Pro and K2 initialize on the steel control point base, which provides the reference geometry. The PortalCam uses the supplied tripod. Uneven ground causes initialization failure or increased mapping thickness.
2. Clear the area. No person, bag, or vehicle in front of the LiDAR during startup.
3. Aim at a feature-rich area. Avoid open plains, large glass surfaces, and anywhere with many moving objects. Those three are the published failure cases.
4. For RTK, reach Fixed outdoors first. Achieve a Fixed solution in open sky before you pick up the scanner. RTK cannot initiate a fix indoors. Walk at least 33 ft (10 m) while Fixed before starting the route.
5. Wait out the full 15-second static countdown and let the app confirm initialization completed. Do not move during it: motion here causes model tilt or reconstruction failure.
6. Then wait at least 10 more seconds after the point cloud appears before lifting the device. These are two separate waits. Lift slowly, without a sudden movement.
7. Pick a start point you can return to. The final loop closure brings you back here, and the procedure for closing it is on 3.2 Route Planning. Getting it wrong costs the whole session's drift correction.

Walking Speed

Speed sets point density. Walking too fast means fewer LiDAR observations per unit of geometry, weaker visual tracking, and faster drift. There is no real-time alarm for moderate excess speed; the cost appears in the processed point cloud.

L2 Pro and K2

• Outdoor, open and feature-rich: 3.3 ft/s. The ceiling, not the target.
• Indoor, general: 1.6 ft/s. The default for nearly all indoor work.
• Tunnels and long corridors: 1.6 ft/s or below.
• High-detail capture of a specific surface: 1.6 ft/s.
• Past glass or mirrors: 1.6 ft/s, with at least 1 m (3.3 ft) of separation.
• Low-light or featureless: 1.6 ft/s or slower.
• Doorway approach: begin slowing 2 to 3 m (6 to 10 ft) out.

PortalCam

The PortalCam runs at roughly half the speed of the other two devices, everywhere.

• Ceiling anywhere: 1.6 ft/s. Not a target, a limit.
• Indoor, dim light, doorways, corners: 1.0 ft/s.
• High-detail capture: 1.0 ft/s.

Half-speed roughly doubles point density at the cost of session time alone. It also leaves margin to react at doorways, transitions, and reflective surfaces, where full speed leaves none.

Posture: L2 Pro and K2

The two devices share most posture rules but not the tilt limit. The K2 is the tighter of the two, and in RTK mode both are tighter still.

Carrying and Standoff

• Keep the scanner clear of your body. No object blocks more than 50 percent of the LiDAR field of view within 1 m (3.3 ft). Hold extra gear behind the scanner relative to travel.
• Stay at least 0.5 m (1.6 ft) from the subject, and 0.5 to 1 m (1.6 to 3.3 ft) on the K2. Alpine practice is to hold 0.6 to 0.9 m (2 to 3 ft) off walls and never closer than 0.3 m (12 in), where depth data degrades.
• Never push the device suddenly into a small enclosed space such as a cabinet under 0.5 m (1.6 ft) deep. The sudden change degrades the LiDAR solution and causes data drift. Approach slowly or scan it from outside.
• Use the carry handle, not the device body, which can cover sensors.

Turning and Vertical Movement

• Rotate over 1 to 2 seconds. Fast turns blur visual frames and confuse SLAM.
• At corners, keep both walls in view. Approach on the diagonal so the wall you leave and the wall you enter both stay in the field of view through the turn.
• Change height gradually over 3 to 5 m (10 to 15 ft). No more than 40 degrees of viewing-angle change at any one moment.

Posture: PortalCam

The PortalCam tolerates far more tilt than the L2 Pro and K2, and far less speed. Its four-camera array covers steep angles the other two cannot.

Lighting Sensitivity

All three devices share the same Multi-SLAM system, but the PortalCam's 3DGS output depends more heavily on image quality than point cloud output does. Lighting that produces acceptable LiDAR data can still produce degraded 3DGS.

• Avoid sudden light-level changes mid-scan. Pause briefly at transitions to let exposure settle.
• Select the low-light scan mode for any route that runs bright to dim and back. It is the published answer for mixed-lighting routes, not just for dark spaces.
• Never point at the sun, windows, or lights. Overexposure degrades tracking and 3DGS color.
• In wind, walk with it. Wind affects PortalCam stability more than the heavier L2 Pro and K2.

Motion Control

The Standing-Still Rule

For LiDAR processing in LixelStudio, only physical movement through space generates new geometry. Standing in place rotates the view but samples the same surfaces from the same vantage point, so after the first few seconds no new useful data arrives. Circle objects at 1.6 ft/s instead of standing in front of them.

The Figure-8 Exception

For 3DGS processing in LCC Studio, controlled figure-8 motions from a near-stationary position add angular variation that improves Gaussian splat density, even though position barely changes. This is the one case where near-stationary scanning adds value. Keep the motion fluid, and use it only when 3DGS output is part of the deliverable.

Smooth Movement

• Plan turns and decelerate into them. Avoid sudden direction changes at full speed.
• Decelerate gradually over the final few feet before a feature you intend to capture in detail.
• Hold posture and height consistent through a session. Switching hold or drifting downward creates trajectory discontinuities SLAM must compensate for.

Session Length and Operator Endurance

Two different limits govern how long you scan, and they are far apart. Do not confuse them.

The Data Limit Comes First

• Plan 12 to 15 minutes per capture. That is the working target and it leaves room for unexpected coverage.
• LixelStudio Map Fusion requires each capture under 20 minutes. A longer capture cannot be fused for point cloud output at all.
• LCC Studio capture guidance is 30 minutes or less per map. Beyond that, reconstruction memory demand becomes the binding constraint.
• A 2-hour capture cannot be fused by either system. Long sessions are split by planning segment boundaries in advance, not by scanning until you are tired and splitting afterwards.

The Fatigue Limit

• Rest 5 minutes every 45 to 60 minutes of working time. Fatigue degrades posture before you notice it, and it degrades scan quality silently.
• Watch for the device drifting toward your waist or your stride lengthening. Both signal fatigue. Stop and rest.
• Past about 2 hours of continuous work, quality drops enough that a rested operator on a fresh set of captures produces better data than pushing on.

Practice Recommendations

• Scan the same space twice, at full speed and at half speed. Compare the point clouds in LixelStudio. The density difference is obvious and makes the speed rule stick.
• Practice doorways and stairwells before client sites. These are the highest-failure transitions and the most expensive to rescan.
• Practice the stop procedure until it is automatic. Returning to the start and standing still before pressing stop is the step most often skipped under time pressure, and it is the one that corrects the whole session. See 3.2 Route Planning.
• Time yourself on known routes. Knowing how long a correct scan of a given area takes is the basis for accurate quoting.