Documentation
User guide
From opening your first scan to issuing the deliverable. Written for the release now in final testing – if something here does not match what you see, tell us and we will fix the page.
Before you begin
PointForge Pro is a 64-bit Windows desktop application. You need an account and an active licence to sign in; a licence is one seat, meaning one person working at a time, on whichever machine they are sitting at.
- Work from a local disk where you can. A fast local SSD makes a real difference to how quickly chunks stream in
- A project can live on a network drive, but copy large source scans locally before editing them
- More graphics memory means more of the project stays resident, so less has to be streamed as you move the camera
Minimum and recommended hardware are not published yet. There is a rendering benchmark built into the application, so the reliable answer for your machine is to run it on your own data.
Install and sign in
- Download the installer from your account page.
- Run it and follow the prompts.
- Start PointForge Pro from the Start menu or the desktop shortcut.
- Sign in with the email address and password you registered on this site.
Seats and sessions
Signing in takes one seat from your licence. Sign out to release it for a colleague, or sign in on another machine and the seat moves with you. Your plan, seat, status and the time of the last validation are shown in preferences, along with a manual refresh and sign-out.
- If a seat is stuck – a machine was reimaged, say – use the seat-recovery workflow to reclaim it
- The session is stored on disk so you are not asked to sign in every time you start the application
If the licence server cannot be reached
An application that is already open keeps working. It is never interrupted mid-session because of a network problem – you can carry on and close it when you are finished.
The next time you start it, if it still cannot reach the server, you will get the sign-in window with a message saying there is no connection to the server. Restore the connection and sign in as usual.
Open your first file
LAS, LAZ, E57 and PLY open directly.
- Use File → Open, or drag the file from Explorer onto the viewport.
- Pick the point cloud. Progress is reported while it loads, and you can cancel.
- Navigate with the mouse – left button orbits, right button pans, the wheel zooms.
- Press F to fit the view to everything visible.
Opening a file from Explorer while the application is already running hands it to the running instance rather than starting a second copy.
Multi-scan E57
- Each scan inside the file appears separately in the Scene Explorer
- Activate, deactivate, show and hide scans independently
- Scan metadata is preserved and shown in the Properties panel
- Embedded images are imported into their own folders and open in the built-in image viewer
IFC reference models
An IFC model can be imported as reference geometry and tessellated for display, so you can compare design intent with the scan in the same workspace. IFC is reference only – PointForge Pro does not author or edit it.
Projects
For anything beyond a look, create a project. A project is a persistent structure that holds your subsets, sections, measurements, groups and meshes, and it keeps your source data intact.
- Choose File → New Project.
- Give it a name and a location – the default projects folder is configurable in preferences.
- Import one or more point clouds as subsets of the project.
- Work. Sections and measurements are preserved between sessions in the project file (
.pdb).
- Several projects can be open at once, in tabs
- Save Project As writes a copy under a new name
- Recent projects are tracked, and entries that no longer exist are validated away
- Search and filter the Scene Explorer when a project gets large
- You are warned about unsaved changes before a project or the application closes
The interface
Viewport
The 3D view in the centre. Everything is drawn here, including section planes, measurements and meshes.
Scene Explorer
The tree on the left. It holds the project hierarchy – subsets, E57 scans and images, sections and their results, measurements and measurement groups, and meshes.
- Tick to show or hide; the active state is separate from visibility
- Active, inactive, visible and hidden items are drawn differently so you can tell them apart
- Multi-select for bulk operations, and right-click for context actions
- Registration status and source provenance are shown against each item
Properties
The panel on the right, showing whatever is selected: file information, header metadata, statistics, geometry and coordinates, spatial reference, and the colour source with its histogram.
Status bar
Along the bottom: memory in use, GPU cache and graphics memory, so you can see what a heavy operation is costing you.
The Scene Explorer and Properties panels can both be hidden when you want the whole window for the viewport.
Reading the cloud
Eight display modes. A feature that is invisible in one is often obvious in another, so switching mode is usually the fastest way to find what you are looking for.
| RGB | Source colour as captured |
|---|---|
| Intensity | Return strength – useful on unpainted surfaces and for reading markings |
| Elevation RGB | Height as a colour ramp |
| Elevation scalar | Height as a scalar field with a scale bar |
| Classification | LAS classes, where the source supplies them |
| Return number | First and last returns – separates canopy from ground |
| Point spacing | Density, for spotting gaps and overlaps in coverage |
| X-ray | Density view, with black and white levels, gamma, strength, curve and inversion |
Histograms and ranges
- Supported modes show an attribute histogram with range controls
- Narrow the visible elevation range to isolate a storey, a carriageway or a stockpile face
- Fine adjustment and a step value for contour-like reading
Colour correction
Brightness, contrast, exposure, gamma, saturation and colour temperature, applied per project or per subset from the Properties panel. Apply when you are happy, or reset to get back to the source values.
Render quality
Four profiles – Fast, Balanced, High Quality and Ultra. Start at Balanced and move in whichever direction your machine argues for.
- Point-splat rendering with adjustable splat size
- Optional accumulation rendering, averaging depth-compatible colours for smoother coverage, with blend radius, surface tolerance and footprint weighting
- A faster mode while the camera is moving, so interaction stays responsive
- Anti-aliasing: 2×, 4×, 8× or 16× MSAA, or FXAA
- Lighting with ambient, diffuse, specular and shininess, and presets – Default, Soft Studio, Dramatic, Outdoor
- Eye-dome lighting, with radius and strength, to bring out edges and relief
- Screen-space ambient occlusion, with radius, strength and sample count
- Compute normals for lighting and surface analysis, visualise them as colour or vectors, and remove them when you no longer need them
Preview budget, sampling percentage and chunk-loading behaviour are all configurable if you want to tune how much the application keeps ready versus how much it streams.
Selecting points
The four tools
- Point – click individual points
- Box (B) – drag a rectangle in screen space
- Brush (N) – paint with an adjustable circular brush; good for organic shapes like foliage
- Polygon – click vertices, then double-click or press Enter to close
Hold Shift to add to the selection and Ctrl to remove from it. The brush also has dedicated add and remove modes if you would rather not hold a modifier.
Selecting only what you can see
Visible-surface filtering applies to box, polygon and brush selection. It keeps only the surface patch nearest the camera, so a box drawn over a wall does not also take the wall behind it. There is advanced tuning for this in preferences if the default is too eager or too cautious for your data.
What you can do with a selection
- Select all visible points, invert the selection, or clear it
- Hide the selected points without deleting them, and restore all hidden points later
- Cut, copy and paste points through the clipboard
- Delete them permanently
- Use it as the source region for a mesh
Selection runs at full resolution using chunk-aware masks, so what you select is what you get – not an approximation taken from the preview.
Editing and transforms
- Move subsets interactively, or apply a numeric shift and scale
- Rotate clockwise or counter-clockwise, and align to the X, Y or Z axis
- Undo and redo cover supported editing and measurement operations
- Each project and subset keeps an edit history
- Clone a subset before trying an alternative approach, and merge subsets into one deliverable
- Reset registration transforms to get back to where you started
Baking
Edits are tracked as project state rather than being written into the point data immediately. When you are satisfied, bake them to commit.
Baking is permanent. If there is any doubt, export to a new file instead – that leaves the project and its history untouched.
Cleaning up
Cleanup operations create a new subset and leave the input alone, so you can compare the two and throw away the one you do not want. Supported operations can be previewed before you apply them.
Reducing point count
- Downsample to a target spacing, a percentage, or a target point count
- Minimum-spacing downsampling and voxel-based reduction
- Feature-aware reduction, which scores curvature so edges and corners survive
- Strict minimum-distance cleanup, and random or reservoir sampling
- Deterministic modes when you need the same result twice
- Original-point-only output, if synthetic centroids would be a problem for you
- Anti-banding controls, grid-origin dithering, grid rotation and two-pass multi-grid processing, for voxel-layer artefacts on flat surfaces
- Pre-voxel reduction to speed up work on a very large cloud
Noise and outliers
- Statistical Outlier Removal – set the neighbour count and the standard-deviation multiplier, then preview
- Dynamic Statistical Outlier Removal – range-aware, for scans where noise grows with distance; optional plane-aware filtering
- Isolated-point and small-cluster cleanup, with preview
- Smoothing, with configurable processing controls
- Several subsets can be targeted at once where the operation supports it
Registering scans
Bring several subsets into one coordinate frame, and see the numbers before you commit to them.
- Choose the fixed reference subset – the one that will not move.
- Choose one or more moving subsets.
- Set the global initialisation: voxel size, and the roll, pitch and yaw search ranges. Wider ranges cost time; narrow them if you already know roughly how the scans sit.
- Run it, then read the per-subset result table.
- Preview the result in the viewport, then apply – or reset and try again.
Reading the results
The table reports status, RMS, overlap, confidence, translation, rotation and any messages, per subset. Acceptance thresholds on RMS and overlap are configurable, so a result that does not meet your standard can be rejected automatically.
Registration reports RMS, overlap and confidence so that you can judge a result. How accurate it is depends on your source data quality, the overlap between scans, your configuration and your own validation. Treat the numbers as evidence, not as a certificate.
Tuning
- Coarse and fine angular search steps
- Candidate ranking, hypothesis limits, overlap thresholds and early stopping
- Optional full sampling, or a sampling limit for speed
- Multi-threaded, with a configurable thread count
- Multi-level refinement, with correspondence distance and trim ratio
- Robust optimisation including Huber weighting and damping
- Rotation and translation stability limits, and convergence thresholds
- Normal-estimation and fallback controls
Measuring
Distance
- Pick the measurement tool.
- Click the first point.
- Click the second point.
- Read the result, and pin it in the scene if you want it to stay while you carry on.
Alongside the direct distance you get the components and projections:
| ΔX | Easting component |
|---|---|
| ΔY | Northing component |
| ΔZ | Height component |
| XY, XZ, YZ | Distance projected onto each plane |
The rest of the toolkit
- Point-to-plane perpendicular distance – for checking flatness and deviation
- Three-point angles, with projections onto each plane
- Polygon area from any number of points
- Convex-hull volume from any number of points
- Detected plane area
Picked points can snap to a local best-fit plane, with a search radius you control, which keeps a measurement off a single noisy point. Input points can be edited afterwards, and area and volume capture has finish, undo-last-point and restart controls.
Planes and corners
- Pick seed points on the surface you want. The plane grows from them automatically.
- Check the transparent overlay – it shows you exactly what was detected.
- Adjust the point-to-plane tolerance if it grabbed too much or too little.
Seed-local fitting follows gentle curvature, and a thin-slab restriction stops a selection jumping to a nearby parallel plane. There are neighbourhood-size, minimum-point and valid-seed thresholds, noise and RMS warnings, seed-normal consistency filtering, parallel-layer splitting, and GPU cleanup passes for speckles and small islands.
For a corner, select each side of it separately – using the point, brush, polygon or box tool – and the intersection line is calculated from the two planes. Its extension, style, thickness and colour are all configurable.
Keeping measurements organised
- Rename measurements with names that are unique in the project
- Group them, move them between groups, remove them from a group, rename or dissolve a group
- Activate, deactivate, show and hide individually or by whole category
- Move labels so they do not sit on top of each other
- Clear pinned labels when the scene gets crowded
- Everything persists in the project, including grouping
These are measurement and inspection tools. They are not certified metrology.
Sections
- Create a section – axis-aligned by X, Y or Z, a custom oriented plane, a box, or a cylinder.
- Position it in the viewport. Picked points can be fitted to a local best-fit plane.
- Set the extraction depth – how thick a slice of cloud the section takes.
- Generate the results: extracted points, and polylines.
Line generation
Corridor width, opening thresholds, closure, minimum chain length and further quality controls determine how the linework is traced from the extracted points. Expect to adjust these once per kind of data, then leave them alone.
Managing sections
- Show or hide plane fill and edges; configure active and inactive plane colours, opacity, size, edge style and thickness, and the highlight colour
- Show, hide or suppress a section – suppression keeps the generated results but stops the clipping and extraction
- Results are tracked as current, stale, empty, busy or in error, and stale results can be regenerated
- Section definitions and their state persist in the project
- Clear all sections at once when you want a fresh start
DXF export
- Export a single section
- Batch export every eligible section
- Export all sections into one DXF, with a layer per section so they stay separable in CAD
- Or one DXF per section, into a folder you choose
- Hidden and suppressed sections can be included in a batch, as long as their results are current
Meshes
- Select the region you want meshed, or select whole subsets if you want all of them.
- Check the analysis – source point count and measured spacing are reported before anything runs.
- Set the target triangle edge length in millimetres. This is a physical size, not an abstract quality slider.
- Read the estimated triangle range and the peak working-memory requirement, then start.
Open or closed
Surface intent can be chosen automatically, or set explicitly. An open surface – a façade, a floor, a road – uses an architecture-oriented local reconstruction. A closed surface uses a screened-Poisson path. Structured and unstructured sources take different analysis paths, and planar reconstruction is used where the evidence supports it.
Cleaning up a result
- Optional healing, applied to a derived copy so the validated base reconstruction is preserved
- Optional small-hole patching, up to a maximum hole diameter you set
- Validation and canonicalisation before the mesh is published into the project
- Source vertex colours retained on supported paths
Simplifying
Simplification creates a linked copy and leaves the original mesh alone. Set a target reduction percentage and a maximum geometric deviation in millimetres – the deviation limit is the one that matters, because it bounds how far the simplified surface is allowed to move away from the original. Boundaries, openings and sharp features are preserved.
Managing meshes
- Meshes are named project entities, with source linkage and build metadata
- Rename, activate, deactivate, show, hide, delete, and inspect properties
- Open the build log and summary for any mesh from the Scene Explorer
- Export as binary glTF (
GLB) or Wavefront (OBJ), with companion material data where applicable
Exporting
Point clouds
Export active project content to LAS, LAZ, E57 or PLY. Before writing, you get validation, any warnings, an estimated file size and a point-count summary.
- LAS 1.2, 1.3 or 1.4, with a compatible point-data format
- Available and missing attributes are listed, so you know what will not survive the format you picked
- Source scale, offset, VLR and format information preserved where full-fidelity export applies
- Coordinate scale and offset automatic or manual
- WKT coordinate-reference information preserved, omitted or supplied by you
- Legacy LAS return-number limits handled safely, either by blocking or by an explicit clamp
- E57 scan name and description, plus selectable fields such as normalised intensity, 16-bit RGB and derived return index and count
- PLY RGB and intensity fields, where available
| Format | Bytes | Contents |
|---|---|---|
| 0 | 20 | XYZ, intensity, returns, classification |
| 1 | 28 | Format 0 plus GPS time |
| 2 | 26 | Format 0 plus RGB |
| 3 | 34 | Format 0 plus GPS time and RGB |
| 6 | 30 | LAS 1.4 extended returns, GPS time |
| 7 | 36 | Format 6 plus RGB |
| 8 | 38 | Format 7 plus near infrared |
All LAS point record formats 0 to 10 can be read on import.
Images
- PNG, JPEG, TIFF or BMP, at a scale you choose
- Current background, or transparent
- Include pinned dimensions and vector labels
- Lock the vector-label scale so annotation stays the same size as you raise the output resolution
- Exported point size is controlled independently of the viewport
Shortcut: Ctrl+Shift+E
PDF and reports
- Viewport to PDF – A4 or Letter, portrait or landscape, four margin presets, fit, fill or 1:1 scaling, white or best-effort transparent background
- Optional header with a custom title and the date and time, optional footer with a page number
- Open the PDF automatically once it is written
- A measurement report PDF carrying measurement names, kinds, values, grouping and project context, with the resolution and status recorded at the time the report was made
PDF is print and export output. There is no direct printer integration.
VR and input devices
VR
VR is optional and needs a compatible OpenXR runtime and headset. It is included in every licence – you supply the hardware.
- Connect the headset and start its runtime.
- In PointForge Pro, start the VR session from the View menu.
- Put the headset on.
There is VR-specific rendering, preview and near-dense detail handling, so close inspection inside the cloud is the point of it rather than a novelty. If a session will not start, check that the runtime is actually running and that the headset is recognised by it first.
3Dconnexion SpaceMouse
Supported for navigation, if that is how you prefer to move through a model.
Sony DualSense
Supported for first-person navigation: sticks, analog vertical movement, view shortcuts, projection and colour controls, and touchpad adjustment of elevation and point size. Deadzones, movement speed, look speed, vertical speed and axis inversion are all configurable, and there is a live view of connection, input, frame rate and diagnostics.
Keyboard shortcuts
These are the shipped defaults. Shortcuts are customisable, with conflict detection, a reset control, and some actions locked – so if one of these does not respond, check whether it has been reassigned in preferences.
Selection
| Box select | B |
|---|---|
| Brush select | N |
| Select all | Ctrl+A |
| Invert selection | Ctrl+I |
| Deselect all, or cancel the tool | Esc |
Editing
| Undo | Ctrl+Z |
|---|---|
| Redo | Ctrl+Y |
| Copy to clipboard | Ctrl+Shift+C |
| Delete selected points | Del |
| Close the active project | Ctrl+W |
View
| Fit view to visible points | F |
|---|---|
| Top view | 1 |
| Front view | 2 |
| Right view | 3 |
| Isometric view | 0 |
| Export image | Ctrl+Shift+E |
Troubleshooting
The view is slow to respond
- Drop the render profile – Balanced, or Fast while you navigate
- Reduce point size, and turn off accumulation rendering and ambient occlusion
- Turn anti-aliasing down, or off
- Deactivate subsets you are not working on rather than only hiding them
- Close other applications competing for the GPU, and update your graphics driver
- Run the built-in benchmark to see which profile your machine is actually comfortable with
Memory is running high
- Close projects you are not using – several open at once all cost something
- Deactivate subsets, scans and image folders you do not currently need
- Lower the preview point budget and the sampling percentage
- Turn off the option to load all chunks, so only camera-selected chunks are streamed
- Watch the status bar while you work: it shows memory, GPU cache and graphics memory
An E57 file will not load
- Check the file is not truncated or corrupt by opening it in another viewer
- Make sure there is enough free disk space for temporary files
- Send us the E57 header information if it still fails – that is usually enough for us to identify the cause
LAS or LAZ coordinates look wrong
- Check the scale and offset values in the file information
- Check the spatial reference – some scanners write local coordinates, which is not an error, just unexpected if you were expecting a projected system
- Large coordinates are handled by rebased rendering, so very large values are not in themselves a problem
Points are black, or nothing appears
- Switch colour mode – try Intensity if RGB is black, since the file may carry no colour
- Check the elevation range if you are using an elevation mode: a narrow range hides everything outside it
- Check that the subset is both active and visible in the Scene Explorer
- Increase point size
- Press F – the camera may simply be somewhere else
It will not start, or crashes on startup
- Update your graphics driver – this is the most common cause
- Try starting it as Administrator
- Reset the settings by deleting
%AppData%\PointForgePro\settings.json
It will not sign in
- If the message says there is no connection to the server, check the machine's own internet access first
- If it says there are no seats available, someone else may be signed in on that licence – sign out there, or use seat recovery
- If the licence has expired, the seat is released; buy another term and sign in again
- Preferences shows your plan, seat, status and last validation, and has a manual refresh
Getting help
Write to support@pointforgepro.com. Your question is answered by someone who works on the software.
Two things make a support request much faster to resolve:
- The log files, from
%AppData%\PointForgePro\logs– application logs and diagnostic views are also reachable from the Help menu - What you were doing, and with which file – the format, roughly how big it was, and which operation was running