Point Cloud Alignment and Coordinate Systems

Point Cloud Alignment and Coordinate Systems

A registered point cloud is a single, consistent 3D dataset, but that does not tell you where it sits. Unless someone decides otherwise, the cloud keeps the origin and orientation of the first scanner position: X and Y start wherever the tripod stood, the Y axis points wherever the scanner happened to face, and zero elevation is the height of the scanner head. That is fine for viewing the data. It is not fine when the cloud has to line up with a site survey, an architect’s existing drawings, a civil model or a Revit model that other consultants will link.

Point cloud alignment to a coordinate system is the step that answers “where is this cloud in the project and in the world?”. It is different from registration, which joins individual scans to each other; we explain that step in how point cloud registration works. This article covers what comes after: choosing a coordinate system, tying the cloud to control points, keeping units and elevations consistent, placing the cloud correctly in Revit, and avoiding the errors that make a model end up in the wrong place.

Registration vs Alignment

The two steps are easy to confuse because both are often described as “aligning scans”.

  • Registration answers: do all scans fit together? Its result is one cloud whose internal geometry is consistent.
  • Alignment, or georeferencing when real-world coordinates are used, answers: where does that cloud sit? Its result is the same cloud moved and rotated into a chosen coordinate system.

A cloud can be perfectly registered and still be in the wrong place, rotated by a few degrees or set at the wrong elevation. The reverse is also true: correct coordinates on a few targets do not fix a poorly registered cloud. Both steps need to be checked separately.

Three Kinds of Coordinate Systems

Most projects use one of three systems. The choice should be made before scanning, because it decides whether survey control is needed on site.

SystemWhat defines itTypical useWhat to watch
Scanner (arbitrary)Origin and orientation of the first scan positionQuick documentation, a single room or object, data that will not be combined with other filesNothing else will line up with it; every consultant has to move it manually
Local or project gridA point chosen on the building (grid A/1, a column center, a corner) and an axis along a main wall or grid line; elevation such as finished floor = 0 or 100 ftRenovation and interior projects where the cloud must match existing drawings or a design modelThe origin and axes must be written down and agreed by everyone; “the corner” can mean different things
State plane, UTM or another mapping systemOfficial projected coordinate system plus a vertical datum (for example NAVD88)Sites, campuses, new construction next to existing buildings, projects shared with civil engineers and surveyorsLarge coordinate values, grid vs ground distances, units (US survey foot vs international foot), vertical datum

Many projects use a combination: the Revit model works in a local project grid, and that grid is tied to state plane coordinates through a known point and a rotation. Revit supports this directly with its Survey Point and Project Base Point, discussed below.

How a Point Cloud Is Tied to Control Points

To place a cloud in a known coordinate system, the cloud needs to contain points whose coordinates in that system are known. These are control points.

Who provides the control

On projects that need official coordinates, control is established by a licensed surveyor, usually with GNSS receivers and a total station. The surveyor sets marks on site and provides their coordinates in the agreed system, with the datum and units stated. On interior projects, control can also come from existing reference points in the building or from an existing CAD or BIM model of the same building.

What happens on site

Targets (checkerboards or spheres) are placed over or next to the control marks, so that the scanner captures them with good geometry. Control should surround the scanned area, not sit in one corner, and it should appear at different heights where elevation matters. We describe how this fits into the field routine in our laser scanning workflow from site to point cloud.

A minimum of three well-spread control points is needed to define position and rotation in 3D. In practice, more points are better: the extra ones act as checks. A good habit is to keep one or two control points out of the transformation and use them only to verify the result. If the cloud lands correctly on points that were not used to compute the alignment, you can trust the result much more than a simple software report.

What happens in the office

In registration software, such as FARO SCENE, the target centers in the cloud are matched to the surveyed coordinates. The software computes the best-fit transformation and reports residuals: how far each target lies from its surveyed position after alignment. Large or uneven residuals usually mean a wrong point ID, a disturbed mark, a units mix-up, or a target captured at a poor angle.

How accurate is the alignment

The cloud’s internal geometry depends on the scanner and registration; our FARO scans have a typical accuracy of ±2–3 mm. How well the cloud fits the coordinate system also depends on the quality of the control itself, the number and spread of control points and how they were captured. The two figures should be reported separately. For more on the measurement side, see 3D laser scanning accuracy.

When control is not needed

Not every project needs survey control. A cloud used for an interior fit-out, a single floor of an existing building or a piece of equipment can be aligned to a local project grid without any surveyed points. The question to ask is simple: will this data ever be combined with a site survey, a civil model or a file in real-world coordinates? If yes, plan control before the scanning visit, because adding it later usually means returning to site.

State Plane Coordinates: Practical Traps

State plane coordinate systems are the usual choice for US projects that involve a site survey. They work well, but they bring a few details that cause most alignment errors. For site work where the cloud also produces terrain and surface data, see our laser scanning topographic surveys.

Grid distances and ground distances

A state plane system projects the curved earth onto a flat grid. As a result, a distance measured on the grid is slightly different from the same distance measured on the ground. The difference is described by a combined scale factor, which depends on location and elevation. A laser scanner measures ground distances, and so does anyone with a tape measure in the finished building.

If a cloud is placed in grid coordinates without considering this, distances in the cloud may no longer match distances on site. The usual solutions are to work in a local ground system tied to state plane through a known point, rotation and documented scale factor, or to agree with the surveyor exactly how the scale factor is applied. Either way, the choice should be written into the project’s coordinate notes.

US survey foot vs international foot

Older state plane coordinates in many states are expressed in US survey feet, while most software uses the international foot. The two differ by about two parts per million. That sounds negligible, but state plane coordinates are often in the millions of feet, so the wrong foot can shift the whole dataset by a few feet. Always confirm which foot the survey uses.

Vertical datum

Horizontal coordinates are only half of the system. Elevations can refer to an official vertical datum such as NAVD88, to an older datum, to a site benchmark or to a project elevation such as first floor = 100 ft. GNSS receivers measure heights above the ellipsoid, which differ from orthometric (sea-level based) heights until a geoid model is applied. A model that sits at the right X and Y but several feet too high or too low is a common result of mixing these.

Units and Formats

Unit errors are simple but frequent. A few rules help:

  • State the units of every file: meters, international feet or US survey feet.
  • Remember that formats differ. E57 stores coordinates in meters; CAD and Revit files may be set up in feet and inches.
  • Check one known distance after every import. If a wall that should be 30 feet long measures about 9 or about 98, the file was read in the wrong units.
  • Keep coordinate order clear. Surveyors often list northing before easting, while most software expects X (easting) first. Swapped values put the cloud in the wrong place without any error message.

Point Clouds in Revit: Survey Point, Project Base Point and Internal Origin

Revit has three reference points, and most point cloud placement problems come from mixing them up. Our scan to Revit services use the same logic on every project.

PointWhat it isHow it is used with point clouds
Internal OriginRevit’s fixed internal zero; it cannot be movedGeometry should stay reasonably close to it; Revit warns when elements are placed very far away, and graphics and precision can suffer
Project Base PointThe project’s own reference, usually a building corner or grid intersectionDefines project coordinates: dimensions and spot coordinates relative to the building
Survey PointA real-world reference, such as a control point with known state plane coordinatesDefines shared coordinates, the system that lets Revit, CAD and point cloud files line up

Linking the cloud

When a point cloud (RCP or RCS) is linked into Revit, the positioning option decides where it lands:

  • Center to center places the cloud’s center at the model’s center. It is useful only for a quick look; it ignores all coordinates.
  • Origin to internal origin places the cloud’s own origin at Revit’s internal origin. This works when the cloud and the model were both set up around the same local zero.
  • By shared coordinates places the cloud according to the model’s shared coordinate system. This is the right choice when the cloud is in state plane or another real-world system and the Revit model’s Survey Point has been set to match.

A typical workflow for a georeferenced project is: the surveyor provides control in the agreed system; the cloud is aligned to that control during processing; the shared coordinates of the Revit model are set from the same control, often by linking a CAD file of the control points and acquiring its coordinates; and only then is the cloud linked by shared coordinates. Modeling starts after a check that the cloud and the control lie on top of each other in plan and in section.

Keeping geometry near the internal origin

State plane coordinates are large numbers. If a cloud with such coordinates is linked origin to internal origin, the building may end up miles from Revit’s zero. The solution is not to move the building far away, but to keep the model geometry near the internal origin and let the Survey Point carry the real-world offset and rotation. Shared coordinates exist exactly for this purpose.

IFC export

When the model is delivered as .ifc, the coordinate setup travels with it if export settings are chosen correctly: the coordinate base (shared coordinates or project base point) and, in newer versions, the coordinate reference system. Open the exported file in a viewer and check that it lands where expected. We explain the format itself in IFC model file format explained.

Common Alignment Errors

ErrorTypical symptomHow to check
Cloud linked center to center or origin to origin by mistakeCloud and model do not overlap at all, or overlap only by chanceCompare known control points in the cloud with the Survey Point and with the surveyor’s list
Wrong unitsBuilding is about three times too large or too smallMeasure a known distance right after import
US survey foot vs international footWhole dataset shifted by a few feet, shape correctConfirm the foot definition in the survey notes
Northing and easting swappedCloud lands far away or mirrored across the siteCheck coordinate order of the control list
Wrong vertical datum or project elevationPlan position correct, elevations off by a constant valueCompare floor elevations with a known benchmark
Grid vs ground scale not agreedSmall, growing length differences on long buildings and sitesCompare a long measured distance with the grid distance
Disturbed or misidentified control pointOne large residual, the others smallRecheck point IDs; use check points not included in the fit
Survey Point moved while clipped or unclipped by accidentShared coordinates change unexpectedly between model versionsRecord base point values and check them at each issue

Most of these errors share one cause: the coordinate decisions were made by different people at different times and never written down. A one-page coordinate note at the start of the project prevents most of them.

What Should Come with an Aligned Point Cloud

When you receive a point cloud from us or from any scanning provider, or send your own cloud for modeling, the files should come with a short description of the coordinate setup:

  • Coordinate system name, including zone for state plane, and the horizontal units with the foot definition.
  • Vertical datum or project elevation, and what zero means on this project.
  • Local origin and rotation if a project grid is used, with its relation to real-world coordinates.
  • Scale factor approach: grid or ground.
  • Control point list with IDs and coordinates, and which points were used as checks.
  • A registration and alignment report with residuals.
  • File formats delivered (for example E57, RCP) and the Revit positioning option to use when linking.

If any of these are unknown, the cloud can still be used, but the uncertainty should be clear before modeling starts. We check these items when clients send us existing data; see what makes a point cloud BIM-ready for the full list of what we look at.

FAQ

What is the difference between point cloud registration and alignment?
Registration joins individual scans into one consistent cloud. Alignment, or georeferencing, moves and rotates that whole cloud into a chosen coordinate system, such as a project grid or state plane coordinates. A cloud can be well registered and still be placed incorrectly, so both steps are checked separately.
How many control points are needed to georeference a point cloud?
At least three well-spread control points are needed to fix position and rotation in 3D. More are recommended, so that some points can be kept out of the calculation and used only to check the result.
Can a point cloud be aligned without a surveyor?
Yes, to a local project grid. The cloud can be aligned to a building corner, grid intersection or existing CAD or BIM model. If the data has to match official coordinates or a site survey, control from a licensed surveyor is the reliable way to do it.
Should I link a point cloud in Revit by shared coordinates or origin to origin?
Use shared coordinates when the cloud is in a real-world system and the model’s Survey Point is set to match. Use origin to internal origin when both the cloud and the model were set up around the same local zero. Center to center is only for a quick visual check.
What is the difference between the Survey Point and the Project Base Point?
The Project Base Point is the project’s own reference, usually a building corner or grid intersection, used for project coordinates. The Survey Point is a real-world reference, usually a control point, and it defines the shared coordinates that let Revit, CAD and point cloud files line up.
Why is my point cloud in the wrong place in Revit?
The most common reasons are the wrong positioning option when linking, mixed units, a different foot definition, swapped northing and easting, or a different vertical datum. Comparing a few known control points in the cloud with the surveyor’s list usually shows which one it is.
Can you work with a point cloud that is not georeferenced?
Yes. We can align it to a local project grid, to existing drawings or to control points you provide, and we agree the coordinate setup in writing before modeling starts.

Decide Coordinates Before You Scan

Coordinate problems are rarely technical in the end. They come from decisions that were not made early enough or not shared with everyone who uses the data. Agree on the coordinate system, units, vertical datum and control before the scanning visit, record them in a short coordinate note, and check the cloud against control before modeling starts. Then the point cloud, the Revit model, the CAD files and the site survey all stay in the same place. ScanM2 travels to sites in any US state for FARO laser scanning and also works from clients’ existing point clouds, with Revit and .ifc delivery set up in the coordinate system your project needs. Learn more about our scan to BIM services, or send us your project details to discuss the coordinate setup.

We work in cities all over the world