8.2 Single Scan Processing

Project Setup

A single-scan project processes one continuous scan session into a point cloud. This is the standard workflow for any scan that fits within your workstation's RAM capacity. For scans that exceed your RAM ceiling, split into segments and use 8.3 Map Fusion.

1. Transfer the raw scan folder from the device to your workstation's NVMe SSD. Copy the entire folder structure. Do not rename, move, or delete individual files within the folder. Verify the transfer completed by comparing folder sizes.
2. Open LixelStudio and create a new project from the Home Page. The Home Page is now integrated into the main application, enabling seamless switching between projects without repeatedly browsing local files. From the Home Page, select Create New Project and point it at the transferred scan folder. Save the project file to a non-system drive (D: or similar). Saving project files to C: may cause disk space bottlenecks or file access conflicts during processing.
3. Verify Windows Power and Sleep settings. Go to Settings > System > Power & Sleep. Set both Screen and Sleep to Never. A sleep interruption during SLAM optimization corrupts the project and requires restarting from import.
4. Confirm LixelStudio is installed and current. LixelStudio supports current device firmware including L2 Pro Drone Mode. If processing fails after a recent firmware update, verify your LixelStudio version is current before troubleshooting further.

SLAM Mode Selection

LixelStudio processes each scan using one of three SLAM modes. The default mode (None) provides the highest accuracy. The other two modes are Special Modes, found in Advanced Settings, used only in specific failure or environment scenarios.

None (Default)

Standard processing mode. Use this for the initial processing attempt in all environments. It provides the highest accuracy in scenes with adequate geometry, stable motion, and sufficient LiDAR return. Most scans should succeed in this mode.

Robust Mode (Off by Default)

Enable Robust Mode from Advanced Settings after a None mode failure, particularly when the processing report shows "LIO trajectory drift." Robust Mode improves success rate in partially degraded scenes and high-motion conditions but may reduce accuracy compared to None mode. Do not use Robust Mode as a first choice. It is a recovery step.

Narrow Scene Mode

Reserved for tunnels, mine shafts, and long corridors. Effective within approximately 1,640 ft (500 m). For corridors over 500 m, segment the scan every 500 m with control points at boundaries and use Map Fusion. Using Narrow Scene Mode in a standard interior or exterior environment will produce a processing failure.

Advanced Settings

These settings are configured before starting the SLAM run. They cannot be changed after processing begins without starting over. LixelStudio streamlined the project processing settings interface for clearer organization of these options.

High-Precision Optimization (New)

The High-Precision Optimization toggle is one of the most useful additions for indoor work without absolute coordinates. When enabled, point cloud leveling accuracy is automatically corrected to a well-aligned horizontal and vertical structure, reducing the need for manual adjustment after processing.

Enable this option for indoor structured scenes processed in relative coordinates (no RTK, no GCP). For outdoor scans or any project using RTK, PPK, or GCP for coordinate transformation, the existing georeferencing pipeline handles leveling and this toggle is not needed.

Filter Level

Controls point cloud noise reduction. Three levels: Weak, Normal, and Strong. LixelStudio upgraded the filtering algorithm with AI-powered noise removal that more effectively preserves fine structures such as railings and pipes. For complex scenes like pipe corridors and railing structures, a dedicated enhancement process now removes trailing noise between structures without damaging geometry.

Normal is appropriate for most environments. For low-density scans (sparse geometry, long-range captures), use Weak. Strong filtering on low-density scans removes legitimate detail along with noise, particularly thin structures. The AI-powered filtering reduces this risk but does not eliminate it entirely.

Point Cloud Enhancement

Generates a denser, more uniform point cloud at the cost of processing time. Powered by the LixelUpSample algorithm, which LixelStudio uses in combination with the upgraded coloring pipeline to improve point density and detail preservation. Available at 5 mm spacing for both L2 Pro and K2, and at 1 mm spacing for L2 Pro only. The 1 mm mode requires 64 GB or more of RAM and substantial free disk space. Enhanced point clouds are segmented into chunks during processing. If a chunk fails, only that chunk needs to be reprocessed, not the entire scan.

Point Cloud Segmentation

Splits the output point cloud into files of a specified size. Default is disabled. When enabled, the default split size is 1 GB, adjustable by the user. Useful when downstream software has file size limitations or when the full point cloud would exceed available viewer memory.

Dynamic Object Removal

Removes point cloud noise from objects that were moving during the scan, such as people, vehicles, or equipment. Effectiveness depends on the original scan collection quality. Following the field technique guidelines in Module 3 produces cleaner results than relying on this filter to compensate for a crowded scan environment.

SLAM Mapping End Time

Allows you to select a specific end time for the trajectory, discarding data after that point. This is useful when the end of a scan contains a drift event or was recorded while the device was being packed up. The selected mapping time must exceed 3 minutes. For RTK or PPK scans, re-enter the positioning settings after adjusting the end time to verify sufficient satellite data remains in the trimmed window. For GCP scans, confirm all control points still fall within the selected time range, with a minimum of 3 points required.

Running the Process

After configuring all settings, start the SLAM optimization. Processing time is approximately 20 to 30 times the scan duration on Recommended-tier hardware with NVMe SSD. A 10-minute scan takes 3 to 5 hours. A 20-minute scan takes 7 to 10 hours.

1. Confirm Power and Sleep settings are set to Never. This is worth checking twice.
2. Close all other applications that consume significant RAM or CPU. Web browsers, Revit, and other processing software compete for memory and can trigger out-of-memory crashes mid-run.
3. Start SLAM optimization. LixelStudio displays an estimated processing time. This estimate is a reference only and varies based on scene complexity, hardware, and concurrent system activity.
4. Do not use the workstation for other intensive work during processing. Memory-intensive tasks launched during the run can cause the SLAM process to crash. Light tasks (email, file management) are generally safe.
5. After SLAM completes, proceed to coloring if camera data is available. LixelStudio substantially improved coloring quality, especially in low-light environments. Coloring requires CUDA 11.6+ and an NVIDIA GPU from the 20-series or newer. The Basic GPU (RTX 2060) satisfies this requirement. See 8.1 Pipeline Overview for full requirements.

Failure Modes and Recovery

When processing fails, the error message and context narrow down the cause. This table covers the failures most commonly encountered in single-scan processing.

Coordinate Transformation

If the scan includes RTK, PPK, or GCP positioning data, the coordinate transformation stage places the point cloud in a real-world coordinate system. LixelStudio redesigned this interface with fast search of common and built-in coordinate systems and clearer naming for easier identification. This configuration must be set before starting SLAM optimization. Changing the target coordinate system after processing requires a complete reprocess from raw data.

RTK

RTK data is applied automatically when present and valid. The RTK status during the scan must have achieved Fixed status with 10 or more satellites, at least 33 ft (10 m) of movement while Fixed, and antenna tilt within 20 degrees. K2 RTK is built-in and standard; L2 Pro RTK uses the optional add-on module. Validation requirements are covered in Module 4: Positioning.

PPK

PPK requires importing RINEX observation files (RINEX 3.0 or later) and configuring the base station parameters. After adjusting any settings, including SLAM Mapping End Time, re-enter the PPK settings to verify sufficient satellite data remains. Full PPK workflow details are in the LixelStudio user manual.

GCP (Ground Control Points)

GCP-based georeferencing requires at minimum 3 control points within the scan. Control point names are case-sensitive and must exactly match the names assigned during field collection. If the SLAM Mapping End Time has been adjusted, confirm all control points still fall within the selected time range.