5.2 Pre-Project Planning for Fusion

Site Assessment

A fusion project that fails in post-processing almost always has its cause in a planning decision made before the first scan began. The site assessment is where you determine whether conditions support the fusion approach you are planning and where you identify the specific locations that will carry or break your result.

Before planning segment breaks, you need to understand the site's spatial organization: how the spaces connect, where the high-feature areas are, where RTK signal will be reliable or absent, and what physical constraints will affect collection. A floor plan walkthrough or site visit specifically for planning purposes pays for itself on every complex project.

Key Questions to Answer During Assessment

• What is the total floor area and estimated scan duration? This determines how many segments you need and whether you are within the 200-minute total limit.
• Where are the logical spatial breaks, floor boundaries, wings, building sections, that make natural segment end points?
• Which areas have rich, stable surface features suitable for overlap zones? Furnished rooms, structured intersections, and equipment areas are good candidates. Empty corridors, stairwells, and glass-heavy areas are not.
• Where will RTK signal be available and where will it not? Outdoor areas and areas near windows may have usable RTK. Deep interior spaces and basements will not.
• Are there any areas with known scanning challenges, reflective surfaces, low light, narrow passages, that need special handling in both field technique and overlap zone placement?
• What is the access situation for each segment? Locked areas, occupied spaces, and time-restricted zones affect the scheduling and order of segment collection.
• Can all segments be collected within 2 hours? Major lighting changes between segments (morning vs. afternoon sun, lights on vs. off, overcast vs. direct sun through windows) degrade the visual feature matching that fusion depends on. Plan to complete all data acquisition within a 2-hour window, or control the lighting environment so it remains consistent across all segments.

Segmentation Strategy

The goal of segmentation is to divide the project into pieces that each stay within processing limits, while ensuring every adjacent pair can be connected through either RTK or control points. You are building a chain, every link must hold.

Sizing Each Segment

Each segment must stay under 20 minutes of scan duration. The 200-minute total limit across a 10-segment maximum means you have 200 minutes of coverage to work with per fusion job. For most commercial projects, segments in the 12 to 18 minute range provide room to manage unexpected coverage needs while maintaining data density per segment.

Plan segment duration conservatively. A segment that ends at 19 minutes has no room to extend if you discover a gap during collection. Targeting 12 to 15 minutes per segment gives you buffer. The 20-minute limit is a hard ceiling, not a target.

Segment Break Point Selection

Segment breaks should happen at locations where you have identified a good overlap zone, an area with rich features where you can end one segment, mark a control point if needed, and begin the next with sufficient shared coverage. Never plan a break at a stairway, narrow corridor, or low-light area. Those are the worst possible places to need alignment geometry.

For multi-floor projects, breaking between floors is natural and usually correct. If a single floor is large enough to require multiple segments, identify a central, feature-rich area as the break point, ideally a lobby, large open office, or equipment room rather than a hallway junction.

Connection Method Decision Per Segment Pair

For each adjacent segment pair, decide before collection which connection method you will use: RTK, control points, or both. Document this in your project plan. The field operator needs to know whether they are relying on RTK validity or physically marking named points at the break location. Discovering on site that RTK is unavailable and no control points were planned is not a recoverable situation.

Overlap Zone Selection

The overlap zone is the shared physical area that two consecutive segments both scan. It must be at least 50 ft (15 m) of coverage area. The quality of the features within that shared area determines whether the fusion algorithm can align the segments reliably.

Characteristics of a Good Overlap Zone

• Rich, varied surface geometry: furniture, equipment, wall features, columns
• Stable and unchanging between the two scan sessions, not an area where people, vehicles, or movable objects will be in different positions
• Accessible from both segment routes without requiring deviation from your planned path
• Well lit with consistent lighting across both segment collection periods. Avoid areas where lighting changes significantly between the first and second segment captures.
• At least 50 ft (15 m) of traversable coverage depth, not just a shared wall or doorway

Locations to Avoid for Overlap

• Stairways and stairwells: narrow geometry, repetitive structure, lighting variation
• Narrow corridors: the scanner cannot maintain the coverage angles needed for reliable feature matching
• Areas with mirror, glass, or highly polished surfaces: reflections create false geometry
• Low-light areas: feature detection degrades, alignment confidence drops
• Areas with high foot traffic during collection: people moving through the overlap zone between segments creates inconsistency
• Areas with complete spatial overlap: one segment covering exactly the same territory as another, with no unique coverage on either side, degrades the algorithm's ability to determine spatial relationships

Control Point Planning

If any segment pair will use control points for connection, including all-indoor projects, the physical locations must be identified during planning, not improvised on site.

Location Requirements

Shared control points must be in the overlap zone between the two segments. Each point needs good visibility in all directions, stable ground surface, and adequate lighting. The device must be placed stably on the ground at the control point location when marking it, not held in the air, not on a soft surface, not on a surface that vibrates from nearby equipment.

Points should be spaced at least 33 ft (10 m) apart from each other within the same segment. Clustering multiple control points in the same small area does not improve alignment quality and can introduce confusion if the algorithm cannot distinguish them spatially.

The Naming Convention

Control point names must be identical across segments down to every character. The fusion algorithm matches points by name. One mismatched character, a capital letter where there should be lowercase, a space, a different number, causes that point to fail. The naming system should be designed to prevent ambiguity, not just to be descriptive.

A reliable convention is short, structured, and unambiguous: "CP01," "CP02," "CP03" is better than "lobby-west-corner" or "point by elevator" because it eliminates variation in capitalization, spacing, and spelling. Document the planned names in your project brief and provide them to the field operator explicitly.

How Many Points Per Segment Pair

The minimum is one shared point per adjacent segment pair. Two is more reliable. For tunnel and narrow corridor projects using Narrow Mode, the requirement is 2 points at each connection, placed at least 16 ft (5 m) apart, with strict position and orientation consistency requirements between the 2 takes. This is covered in detail in the Field Collection section.

Permanent Control Points for Recurring Projects

If the project involves a recurring scan cycle (periodic rescans of the same facility for construction progress, facility management, or asset tracking), the control points planned for the baseline scan are not temporary. They become the registration infrastructure for every future session.

Control points are written into the project file during live scanning. There is no post-processing tool in LixelStudio or LCC Studio that inserts a named control point into existing data. If the baseline scan is collected without named control points in the planned overlap zones, there is no mechanism to fuse future rescans against it using the control point connection method. The baseline would need to be recollected with control points, or the first rescan would need to replace the full dataset rather than merging with it.

For indoor facilities where RTK is unavailable (data centers, hospitals, industrial interiors), control points are the only viable connection method between sessions collected months apart. Plan accordingly:

• Install permanent physical markers at control point locations. Floor paint, adhesive targets, survey nails, or embedded anchors eliminate the need to re-survey positions at each visit. Shared point positions only need to be approximately consistent (within 1.6 ft / 0.5 m horizontal, within 20 degrees orientation, approximately 4 in / 10 cm height), but a physical mark removes the guesswork.
• The naming convention established for the baseline becomes permanent. Every future rescan must use the same names at the same locations, character for character. Document the names, photograph each location, and store the reference material with the project archive.
• Plan the baseline scan with future overlap zones in mind. Identify which areas are most likely to change between sessions and ensure control points bracket those zones so partial rescans can be fused against the unchanged baseline.

Detailed guidance on recurring scan alignment, including permanent control networks, rescan route planning, and residual verification, is covered in 3.5 Data Center Scanning.

Resource Planning

Large fusion projects require more than just planning the scan structure. The logistics of the collection day need to be accounted for explicitly.

Batteries

The L2 Pro and K1 provide approximately 90 minutes of operating time per battery charge, covering 4 to 6 segments at the recommended 15 to 20 minute segment length. For a project requiring 8 segments, plan for 2 battery charges plus one spare. Cold weather reduces runtime by approximately 20%, so in winter conditions budget a third charge and plan charging windows between collection periods.

Storage

The L2 Pro and K1 generate approximately 60 to 80 GB of raw data per hour of scanning. A 200-minute project produces 200 to 270 GB of raw data. The processing machine requires free disk space equal to approximately 5 times the raw data size to accommodate intermediate files during SLAM optimization and fusion. Confirm storage capacity on both the scanner and the processing machine before the project begins.

Personnel

Multi-segment projects benefit from at least one additional person managing documentation: tracking segment numbers against the planned structure, confirming control point names are entered correctly, and noting any deviations from the planned route or overlap zone. The scanner operator cannot simultaneously manage execution quality and project documentation at the level a fusion project requires.

Processing Hardware

Map Fusion requires substantially more processing power than single-scan processing. For fusion jobs and aerial-ground fusion, the recommended hardware is a 16-core processor (AMD Ryzen 9 9950X or equivalent), 64 GB DDR5 RAM with 96 to 128 GB recommended for larger datasets, and an NVIDIA RTX 3090 with RTX 4090 recommended for optimal performance. Processing time estimates run approximately 20 minutes of compute time per minute of scan data at Medium Quality settings. For a 200-minute project, that is approximately 65 hours of processing time on recommended hardware. Plan accordingly.

Naming Conventions for Segments

Consistent segment naming prevents confusion during data management and processing. Official guidance recommends a structured format that identifies the building, floor, and segment letter: BuildingName_Floor1_SegmentA, BuildingName_Floor1_SegmentB, BuildingName_Floor2_SegmentA, and so on.

Whatever convention you use, apply it before collection begins, communicate it to the field operator, and apply it consistently to every segment file from transfer through processing. Segments named scan1, scan2, newscan, finalscan2 on a 10-segment project become unmanageable during fusion setup and impossible to audit afterward.

Pre-Project Checklist

• Total scan duration estimated and confirmed within 200-minute limit
• Number of segments determined and each sized under 20 minutes
• Segment break locations identified at feature-rich, accessible areas
• Connection method confirmed for each adjacent segment pair (RTK, control points, or hybrid)
• Overlap zones identified and confirmed at 65 to 100 ft (20 to 30 m)
• Overlap zones verified to avoid stairs, corridors, low light, and reflective surfaces
• Control point locations identified within each overlap zone
• Control point names defined and documented using consistent, unambiguous convention
• For recurring scan projects: permanent control point locations marked physically and documented for future session re-occupation
• All segments can be collected within a 2-hour window, or lighting environment is controlled to remain consistent across the full collection period
• Processing machine storage confirmed (approximately 5 times raw data size for workspace)
• Processing hardware confirmed against fusion requirements (16-core CPU, 96 to 128 GB RAM, NVIDIA GPU)
• Battery supply planned for total project duration plus cold-weather buffer if applicable
• Segment naming convention defined and communicated to field operator