Part 7: Complete LixelStudio Tool Reference

---

Quick Field Guide

Initial Data Cleanup

Clipping Box (3D Volume Selection):

• Use for: Rectangular boundaries, quick removal of large unwanted areas
• Process: Select data → 3D Clipping → Adjust box (Translate/Rotate/Scale)
• Tip: Clip conservatively first pass, can always refine further
• Critical: Check vertical boundaries (roof/foundation) from multiple viewpoints

Polygon Clipping:

• Use for: Irregular boundaries, surgical precision
• Process: Select data → Clipping → Polygon Selection → Draw boundary → Double-click to close
• Inner Clip: Keep inside, remove outside (isolate subject)
• Outer Clip: Keep outside, remove inside (delete unwanted objects)
• Tip: Choose best view first (Top for footprints, Front/Side for vertical features)

Delete:

• Removes point cloud from current session (file remains on disk if saved)
• Can recover via File → Import Data if file exists

Learn more about cleanup tools →

Data Refinement

Denoising (Remove Scan Artifacts):

• Neighborhood Points: Default 10 (increase for dense data, decrease for sparse)
• Standard Deviation Multiple:
  • 0.5 = aggressive (removes more noise, risks losing real geometry)
  • 1.0 = default balanced
  • 2.0 = conservative (preserves geometry, leaves more noise)
• Always denoise before resampling
• Sources of noise: Dust particles, glass reflections, motion blur, mixed pixels at edges

Resampling (Uniform Point Density):

• Uniform (Voxel) Sampling: Preferred method (creates truly uniform density)
• Voxel Size Guidelines:
  • 5mm = Ultra-detailed (40,000 pts/m²) - heritage, industrial metrology
  • 1-2cm = General architectural (2,500-10,000 pts/m²) - most work
  • 5cm = Large scale/terrain (400 pts/m²) - building exteriors
• Random Sampling: Fast but maintains original density bias (not uniform)
• Always save non-resampled master before downsampling

Smoothing (Reduce Surface Noise):

• Changes your data - use carefully
• Use when: Visual appearance > absolute accuracy (visualization, mesh generation)
• Don't use when: Measurements needed, as-built accuracy critical
• Neighborhood Radius: Larger = more aggressive smoothing
• Tip: Apply selectively (smooth floors, keep walls original) or create two versions

Learn more about data refinement →

Alignment and Positioning

Manual Registration (Align Two Point Clouds):

• Requires: Minimum 3 correspondence points (not on straight line)
• Process:
  1. Select both clouds
  2. Manual Registration → Verify reference vs. target
  3. Pick same feature in both clouds (3-5 pairs)
  4. Register → Check RMSE (low = good)
• Tips: Choose sharp corners, spread points throughout overlap, use multiple viewpoints

ICP Registration (Automated Refinement):

• Requires good starting alignment from manual registration first
• Parameters:
  • Max Iterations: 20 (default works well)
  • RMSE Threshold: 1e-6 typical
  • Voxel Radius: Default (adjust for performance)
• Uses: Local optimization, makes micro-adjustments
• Can fail: If starting alignment poor, converges to wrong solution

Align (Reorient Coordinate System):

• Use when: Need building parallel to axes (easier floor plans/sections)
• Changes coordinates permanently (different from view rotation)
• Process: Align → Draw axis along desired direction → Apply
• Work in: Top view (horizontal), Front/Left (vertical)
• Tip: Verify from multiple views before saving

Merge (Combine Multiple Clouds):

• Requires: Clouds already in same coordinate system
• Process: Select all → Merge → Choose attributes to preserve
• Attributes: XYZ (mandatory), RGB, Intensity, Classification, GPS time
• Tip: Verify alignment before merging (can't easily separate after)

Learn more about alignment →

Navigation and Viewing

Mouse Controls:

• Middle button + drag: Rotate around center
• Right button + drag: Pan
• Scroll wheel: Zoom in/out
• Double-click left: Set rotation center point

Standard Views:

• Front, Back, Left, Right, Top, Bottom (Quick Toolbar buttons)
• Eliminate perspective distortion for measurements

Rotation Modes:

• Turntable: Vertical stays vertical (better for architecture)
• Trackball: Free rotation in all directions

Camera Modes:

• Perspective: Natural depth perception, use for exploration
• Orthographic: No distortion, use for measurements/technical work
• Switch between modes based on task

Rendering Modes:

• EDL (Eye-Dome Lighting): Enhances depth perception, reveals surface detail
• X-Ray: Transparency to see through layers
• Both are display-only (don't affect exported data)

Measurement Tools

Point Measurement:

• Click any point → See X, Y, Z coordinates + RGB + Intensity
• Use for: Exact locations, elevation queries, coordinate verification

Distance Measurement:

• True 3D distance between points
• Continuous mode: Keep clicking to chain segments (perimeters, path lengths)
• Right-click: Undo last point without losing chain

Area Measurement:

• Auto-switches to Orthographic (required for projection plane)
• Click polygon vertices → Double-click to close
• Result: Projected area (not actual if sloped)
• Buffer Width: Thickness for uneven surfaces

Angle Measurement:

• Three points: start, vertex (angle location), end
• Measures 3D angle
• Critical: Precise vertex placement, all points in same plane for planar angles

Multi-Point Measurement:

• Build table of many points with coordinates + labels
• Export: TXT or CSV for use in Excel, GIS, etc.
• Use for: Systematic location extraction (outlets, markers, control points)

Density Measurement:

• Points per unit area in selected region
• Quality assessment: Did you capture enough detail?
• Guides resampling decisions

Learn more about measurements →

Slicing and Profiles

Horizontal Slice (Floor Plans):

• Standard cut height: 1.2m (4 ft) above floor
• Thickness: 20-30cm typical balance
• Process: Horizontal Slice → Front view → Click bottom then top → Adjust → Export
• Step Function: Auto-create multiple floors at regular intervals
• Export: LAS (point cloud) or RCP (Autodesk)

Vertical Slice (Building Sections):

• Work in Top view, define cutting line
• Thickness: 10-30cm typical
• Same interface as Horizontal Slice, different orientation
• Use for: Wall elevations, internal building sections

Profile Analysis:

• Interactive cross-sections with dedicated workspace
• Select rectangular area → Specify buffer width
• Dedicated window: Own view controls + full measurement tools
• Export: LAS, DXF, or orthophoto
• Use for: Detailed examination without full cloud clutter

Panorama Overlay:

• Synchronized photo + point cloud view (requires camera coloring)
• Navigate trajectory like Google Street View
• All measurements work with photorealistic view
• Client presentations: Intuitive for non-technical viewers
• Floor slicing: Separate multi-story buildings by elevation

Learn more about slicing →

2D Drawing and Vectorization

AI-Extract (Automated Wall Detection):

• Parameters:
  • Ground Sampling Distance: Analysis detail level
  • Point Cloud Simplification: Voxel downsampling for speed
  • Snap to 90°: Enable for architecture (Manhattan World)
  • Min Points per Segment: Detection sensitivity
  • Max Extension Length: Corner intersection tolerance
• Workflow: AI-Extract → Review → Manual refinement

Manual Drawing Tools:

• Basic: Line, Polyline, Arc
• Shapes: Rectangle (2-pt/3-pt), Circle (2-pt/3-pt/center)
• Editing: Extend, Break, Merge, Trim, Intersect, Clone, Move, Delete
• Snapping: Midpoint, Endpoint, Intersection, Perpendicular, Center
• Orthogonal/Polar: Constrain to H/V or angle increments

Export:

• Internal save: Preserves work in LixelStudio project
• Export DXF: Clean vectors for CAD software (AutoCAD, BricsCAD, etc.)
• Result: Dimensionally accurate line drawing from point cloud

Data Recording (Annotated Documentation):

• Add text annotations at specific locations
• Capture screenshots with annotations + measurements visible
• Generate report: Title, organization, formatted document
• All-in-one: Annotations, views, dimensions in professional output

Workflow Quick Tips

Typical Processing Order:

1. Clip (remove unwanted areas)
2. Denoise (remove artifacts)
3. Resample (uniform density)
4. Smooth (optional, if visual > accuracy)
5. Align/Register (position correctly)
6. Merge (if combining multiple clouds)

Before Measurements:

• Use Orthographic view (not Perspective)
• Use standard views when possible (Front, Top, etc.)
• Check point density in measurement area

Before Drawing Floor Plans:

• Create horizontal slice first (~1.2m height, 20-30cm thick)
• Try AI-Extract for initial wall detection
• Manually refine + add details

File Management:

• Name clouds descriptively
• Keep originals before major operations
• Export deliverables with clear names

---

Full Explanatory Guide

Initial Data Cleanup: Removing What You Don't Need

Processed point clouds typically contain more data than needed. Scans often include parking lots, neighboring structures, or ground vegetation beyond the project area of interest. Before investing time cleaning and refining your data, remove these large sections that are completely irrelevant to your project.

Quickly reducing your dataset to just the areas of interest saves both time and computational resources in subsequent operations.

Clipping Box: 3D Volume Selection

The Clipping Box allows you to define a rectangular box in 3D space and keep only the points inside that box. This is ideal when your scan includes a large surrounding area but you only need to focus on a specific building or site.

When to Use the Clipping Box

Use the Clipping Box when you have clearly defined rectangular boundaries for your area of interest. For example, if you scanned an entire block but only need one building, the Clipping Box lets you quickly isolate that building. This tool works best for subjects that roughly fit into a rectangular volume, such as individual buildings, rooms, or machinery installations.

The Clipping Box is less effective for irregularly shaped subjects or when you need to preserve complex geometries that don't fit neatly into a rectangular boundary. In those cases, Polygon Clipping provides more precision.

Step-by-Step: Using the Clipping Box

1. Select your point cloud data in the data list on the left side of the screen

2. Click "3D Clipping" in the Tools menu. A blue semi-transparent box appears around your point cloud representing the clipping volume

3. Adjust the box position using three manipulation modes:

   • Translate mode: Moves entire box without changing size (use axis handles to move along X, Y, or Z)
   • Rotate mode: Spins box around its center (useful when building isn't aligned with cardinal directions)
   • Scale mode: Adjusts box size by moving faces inward or outward

4. Switch between viewpoints as you adjust. Use viewport controls (Front, Top, Left, Right views) to check your clipping box from multiple angles. Always verify from at least three different viewpoints before saving.

5. Click "Save" to export the clipped data. LixelStudio creates a new point cloud containing only the points inside your box. Original data remains untouched.

6. Click "Close" to exit the tool

Critical Tips:

• Clip conservatively on first pass - Better to include slightly more than needed rather than accidentally cutting off important geometry
• Pay special attention to vertical boundaries - Leave vertical buffer space to ensure you're capturing complete structure
• Use Reset button if you make a mistake - Faster than trying to undo individual adjustments
• Remember non-destructive workflow - Original data remains available if you need different clip boundaries

Polygon Clipping: Precision Selection

While the Clipping Box works well for rectangular selections, real-world subjects rarely fit perfectly into boxes. The Clipping tool provides flexibility by allowing you to draw custom polygon shapes around your area of interest.

When to Use Polygon Clipping

Polygon Clipping excels for irregular boundaries: stockpiles with organic edges, buildings with complex footprints, removing vehicles or people from street scenes, or extracting specific equipment from dense industrial environments.

The polygon approach provides surgical precision. Unlike the Clipping Box which works in 3D volume, Polygon Clipping works primarily in 2D from your current view, but extends infinitely in the depth direction. When you draw a polygon around an object from the top view, you're selecting a vertical column of points extending from bottom to top of your dataset.

Step-by-Step: Using Polygon Clipping

1. Select your point cloud in the data list, then click "Clipping" in Tools menu

2. Choose your view carefully - Switch to the viewpoint that gives the clearest view of your subject's boundaries (Top view for building footprints, Front/Side for vertical features)

3. Select "Polygon Selection" from the toolbar

4. Draw your polygon by clicking points along the boundary. Each click places a vertex. Use Right-click to undo your last vertex if you make a mistake. Take your time and zoom in to verify accuracy.

5. Close your polygon by double-clicking when you return close to your starting point

6. Choose your clipping mode:

   • Inner Clip: Keep only points inside polygon, remove everything outside (for extracting/isolating subject)
   • Outer Clip: Keep everything outside polygon, remove points inside (for deleting unwanted objects)

7. Click "Save" to execute the clipping operation

Advanced Techniques:

View-based clipping concept: When you draw a polygon in Top view, you're creating a vertical column selection. This is perfect for isolating buildings but might capture more than intended if there are objects at different elevations within your polygon.

Combined operations: Professional workflows often combine multiple clipping operations. First use Clipping Box to limit vertical extent, then use Polygon Clipping in Top view to precisely define horizontal boundaries.

Rectangle Selection tools: Work similarly to Polygon Selection but constrain to rectangular shape. Faster for quick rectangular selections (only need two clicks for diagonal corners).

Data Refinement Tools

After removing unwanted areas, refine your data to improve quality and usability.

Denoising: Remove Scan Artifacts

Noise in point clouds appears as scattered points that don't represent actual surfaces - dust particles captured mid-air, spurious returns from glass reflections, motion blur from camera movement, or mixed pixels at surface edges.

Denoising identifies and removes these outlier points based on statistical analysis. Each point is compared to its neighbors. If a point is significantly farther from its neighbors than expected for a real surface, it's classified as noise and removed.

Parameters:

Neighborhood Points (default 10): How many neighboring points to examine for each point. Increase for dense data (more neighbors available for statistical analysis), decrease for sparse data (fewer neighbors available).

Standard Deviation Multiple:

• 0.5 = aggressive: Removes more noise but risks losing real geometry on thin features or sharp edges
• 1.0 = default balanced: Good starting point for most applications
• 2.0 = conservative: Preserves geometry at the expense of leaving more noise

Best practices:

• Always denoise before resampling - Resampling can "solidify" noise by incorporating it into voxel averages
• Start conservative - Use 2.0 multiplier first, then increase aggressiveness if needed
• Check results - Zoom in to examine thin features and edges to verify they weren't removed

Resampling: Uniform Point Density

Raw scan data has highly variable point density. Areas where you walked slowly have dense coverage while distant areas or quickly-passed sections have sparse coverage. Resampling creates uniform point density throughout your cloud.

Uniform (Voxel) Sampling (recommended): Divides 3D space into a regular grid of voxels (3D pixels). All points within each voxel are replaced by a single point at the voxel center or as an average of the points in that voxel.

Voxel Size Selection:

• 5mm voxels = 40,000 pts/m² = Ultra-detailed for heritage documentation, industrial metrology, detailed mechanical systems
• 1-2cm voxels = 2,500-10,000 pts/m² = General architectural work, most building documentation
• 5cm voxels = 400 pts/m² = Large-scale work, building exteriors, terrain modeling

Critical workflow principle: Always save your original non-resampled point cloud as a master. Resampling is lossy - you cannot recover original density after downsampling. Save the master, then create resampled versions for different purposes.

Random Sampling: Alternative method that randomly selects a specified percentage of points. Faster than voxel sampling but doesn't create uniform density - just reduces overall density while maintaining the original bias pattern.

Smoothing: Reduce Surface Noise

Smoothing changes your data by moving points to create smoother surfaces. This is fundamentally different from denoising (which removes points) or resampling (which changes density).

When to use smoothing:

• Visual appearance is priority over absolute accuracy
• Preparing data for mesh generation
• Creating visualizations for client presentations
• Reducing texture on surfaces you know should be smooth

When NOT to use smoothing:

• Measurements will be extracted from the data
• As-built accuracy is critical
• Legal or regulatory documentation
• Change detection or deformation monitoring

Neighborhood Radius: Controls how many nearby points influence each point's smoothed position. Larger radius = more aggressive smoothing = smoother surfaces but more geometry distortion.

Selective smoothing: Instead of smoothing entire cloud, consider smoothing only specific areas. For example, smooth floors (which should be smooth) while keeping walls original (which may have important surface texture).

Alignment and Positioning

Manual Registration: Align Two Point Clouds

Manual Registration aligns two point clouds by identifying corresponding features in both clouds. This is the foundation for combining scans that weren't captured in the same coordinate system.

When you need manual registration:

• Combining scans from different sessions without shared control points
• Aligning scan data with CAD models
• Creating complete models from partial scans captured from different positions

Critical requirements:

• Minimum 3 correspondence points (4-5 recommended)
• Points must NOT lie on a straight line (creates mathematically underdefined transformation)
• Points should be spread throughout overlap region (not clustered in one area)

Step-by-Step Process:

1. Import both point clouds into LixelStudio

2. Click "Manual Registration" in Tools menu

3. Verify which cloud is Reference and which is Target. The Target cloud will be moved to align with the Reference cloud.

4. Pick correspondence pairs:

   • Click plus button to begin
   • Click on a distinctive feature in Reference window (corner, light fixture, doorknob, etc.)
   • Immediately click the same feature in Target window
   • Repeat for at least 2 more pairs (3-5 pairs total recommended)

5. Choose features wisely:

   • Sharp corners are excellent (precisely defined)
   • Avoid smooth curved surfaces or flat walls (ambiguous location)
   • Spread points throughout overlapping region
   • Ensure points form 3D constellation, not straight line

6. Click "Register" - Software calculates transformation and applies it to Target cloud

7. Examine RMSE (Root Mean Square Error):

   • Small RMSE (under a few centimeters for building-scale) = successful registration
   • Large RMSE = correspondence points weren't picked accurately OR clouds cannot be aligned with simple rigid transformation

8. If unsatisfied, click "Reset" and re-pick correspondence points more carefully

9. Click "Confirm" when satisfied - LixelStudio creates new registered point cloud

Expert tips:

• Spend time examining both clouds before picking points
• Use view controls extensively while picking - verify same location from multiple angles
• Manual registration is typically not the final step - use ICP Registration afterward for refinement
• If clouds have very little overlap or few distinctive features, registration may not be possible

ICP Registration: Automated Alignment Refinement

ICP (Iterative Closest Point) Registration is the automated refinement tool that follows manual registration. While manual registration gets clouds approximately aligned, ICP considers the entire point cloud and makes micro-adjustments.

How ICP works:

1. Starts with current cloud positions (should already be roughly aligned from manual registration)
2. Identifies closest point pairs between two clouds
3. Calculates transformation to minimize distances between these pairs
4. Applies transformation, moving target cloud slightly
5. Repeats until convergence or maximum iterations reached

Critical insight: ICP is a local optimization algorithm. It makes small adjustments to improve alignment but cannot fix major misalignments. Manual registration must come first - ICP needs a good starting point to succeed.

Using ICP:

1. Complete manual registration first - Verify clouds are roughly aligned

2. Click "ICP Registration" with both clouds selected

3. Set parameters:

   • Maximum Iterations: 20 (default works well)
   • RMSE Threshold: 1e-6 typical (stopping criterion when improvement becomes smaller than threshold)
   • Voxel Radius: Default provides good balance (adjust only if performance problems or convergence issues)

4. Click "Confirm" - Algorithm processes (can take several minutes for large datasets)

5. Examine results:

   • Zoom in to overlap areas
   • Verify surfaces align accurately
   • Check final RMSE if reported (few millimeters = excellent for architectural scanning)

Troubleshooting:

• Results worse than manual registration: Starting alignment wasn't good enough. ICP converged to wrong solution. Improve manual registration.
• Very slow or hangs: Point clouds too large. Use clipping tools to reduce to just overlapping regions.
• Surfaces don't align perfectly: Objects may have moved between scans, or there's actual deformation. Perfect alignment may be impossible.

Measurement Tools: Extracting Quantitative Information

Point Measurement

Click on any point to see:

• X, Y, Z coordinates in your project's coordinate system
• RGB color values (exact red, green, blue components)
• Intensity value (return signal strength recorded by scanner)

Common uses:

• Know exact location of a feature
• Report elevation of specific items (manhole covers, corners, equipment)
• Verify whether two points are truly at different elevations
• Verify accuracy of coordinate transformation (compare measured coordinates to surveyed coordinates)

Technique: Click Point Measurement button → Click points in cloud → Each click adds measurement marker → Click button again to exit measurement mode

Distance Measurement

Calculates straight-line 3D distances between points. Accounts for vertical separation if points are at different elevations.

Continuous mode: Keep clicking to measure additional connected segments - perfect for:

• Room perimeters (click each corner around room)
• Path lengths (click at each bend point)
• Total pipe length along route with multiple turns

Right-click: Undo previous point without losing entire measurement chain

Best practices:

• Zoom in close to features being measured
• Click precisely on intended points (not nearby points visually in front)
• Remember: Clean data enables accurate measurements (denoise first)

Area Measurement

Calculates projected area of polygon you draw on point cloud.

Key characteristics:

• Auto-switches to Orthographic projection (essential for area calculations)
• Projected area, not actual surface area (for sloped roof viewed from above, you get horizontal projected area, which is less than actual roof area)
• Buffer Width setting: Thickness for selection on uneven surfaces

Process: Click vertices of polygon → Double-click to complete → Software displays enclosed area

Common uses: Building footprints, zoning calculations, surface area calculations, surveying applications

Horizontal Slice: Creating Floor Plan Data

The Horizontal Slice tool extracts horizontal cross-sections from your point cloud at specified elevations. Think of it as taking a horizontal saw cut through your building at a specific elevation.

These slices are the foundation for creating floor plans. When you slice at standard architectural cut height (typically 1.2 meters or 4 feet above floor), you capture wall cross-sections, door and window openings, and general layout of the space.

Key parameters:

• Slice elevation: Z coordinate where slice is taken
• Slice thickness: How much vertical height the slice includes (20-30cm typical)
• Step length: For creating multiple evenly-spaced slices automatically

Step-by-Step Process:

1. Select point cloud to slice from data list

2. Click "Horizontal Slice" in Tools menu - Interface reorganizes to show three orthogonal views

3. Create first slice in front view:

   • Click once at bottom elevation of desired slice
   • Click second time at top elevation
   • Highlighted band appears representing your slice

4. Adjust precisely using arrow handles:

   • Top/bottom arrows move slice boundaries
   • Yellow circle in center moves entire slice while maintaining thickness

5. For floor plan: Click at ~0.7m (2.5 ft) and ~2m (6.5 ft) to capture walls, openings, furniture while excluding floor and ceiling

6. Create multiple slices using Step Function for multi-story buildings

7. Export: LAS (point cloud) or RCP (Autodesk format)

Best practices:

• Standard cut height: 1.2m (4 ft) above floor for typical floor plans
• Thickness: 20-30cm provides good balance between capturing wall geometry and excluding noise
• Verify from multiple views before finalizing
• Name slices descriptively when saving