When you record control points and press Station, Dusty rotates and shifts the laser tracker's frame of reference so your scanned points land as close as possible to their CAD positions. That process is called best-fit. Anchoring a control point changes the deal: it tells best-fit to prioritize that one location, and to push the leftover error onto the points you didn't anchor.
The two sets of points Dusty compares
Best-fit always works with two versions of the same control points:
| Set | Where it comes from | What it represents |
|---|---|---|
| CAD control points | The coordinates in your CAD file and the CSV you upload | Where each point is supposed to be |
| Scanned control points | What the laser tracker measures when you place the reflector on each marked point | Where each point actually is |
The CAD file and the laser tracker each start in their own coordinate system. Best-fit's only job is to bring those two systems into agreement. Use the legend below for the following illustrations.
What happens when you press Station![]()
- Dusty takes both sets of points. You need at least 3 control points that don't all fall on one line. More than 3 is better.
- It finds the geometric center of each set. The center of your CAD points and the center of your scanned points each become a pivot.
- It slides one center onto the other. This is the shift, or translation.
- It rotates the scanned set onto the CAD set. Dusty solves directly for the single rotation that makes the total leftover gap across every recorded point as small as possible — a standard least-squares fit.
- It reports what's left over. Whatever gap remains at each point after the best possible fit is the error you see for that point in the app.
Because the fit is averaged across every point you record, recording more than the 3-point minimum improves accuracy. Extra points give best-fit more samples of the jobsite and dilute the effect of any single point that's slightly off.
Your layout never gets stretched or shrunk
Best-fit only ever applies two moves: a rotation and a shift. It never scales. No matter how large your control point errors are, your layout prints at true 1:1. Errors can rotate or offset where the layout lands on the slab; they can never change its dimensions.
Dusty also never edits your layout. It solves for where the tracker is standing and which way it's facing — your file is untouched.
What anchoring a control point does
Anchoring locks a point in place. Instead of being one vote among equals in the fit, an anchored point becomes the reference the fit is built around. Behind the scenes Dusty weights that point so heavily that the solution has no choice but to satisfy it first.
That has a cost, and it is the whole point (pun intended) of understanding this feature: anchoring never reduces total error. It moves error off the points you care about and onto the points you don't.
Zero, one, or two anchors: the tradeoff
| Zero anchors | One anchor | Two anchors | |
|---|---|---|---|
| What the fit optimizes | All control points equally | The anchored point first, then the rest | The two anchored points first |
| Rotation pivot | Geometric center of each point set | The anchored point | The midpoint of the line between the two anchors |
| Error at the anchored point(s) | n/a | Effectively zero | Equal at both points — not zero |
| Largest error on the remaining points | Lowest | Higher than zero anchors | Higher than one anchor |
| What it guarantees | Nothing specific — the best average | A position | A position and a direction |
| Use it when | You have good surveyed control | One specific spot must be exact | The layout must align to a line, like an existing wall |
The pattern is consistent: every anchor you add tightens one location and loosens everything else.
Anchoring one point
With one anchor, that point becomes the pivot. Its scanned position and its CAD position are forced to sit on top of each other, so its error goes to zero. The rotation is then solved to fit the remaining points as well as possible around that fixed pivot.
Anchor one point when a single location is the one that has to be right — a stairwell corner, a column intersection, the end of a major wet wall. The layout is guaranteed to land correctly there.
One anchor pins a position, not a direction. That is the catch, and it matters most when you're trying to line up with something that already exists on the floor. One anchor guarantees the layout at that one point and nothing else. The remaining points absorb all the leftover error, which means the layout can come out noticeably skewed relative to the object you were aiming at — and there is no practical way for you to predict how skewed before you print. If your goal is to align to an existing wall, column line, or gridline, use two anchors instead.
Anchoring two points
Two anchors define a baseline — the imaginary line between them. Dusty pivots around the midpoint of that baseline and rotates until the scanned baseline lies on the CAD baseline. Rotation is no longer a compromise across four points; it is set by the baseline itself.
Two things follow from that:
- The two anchored points always end up with equal errors. They sit the same distance from the pivot, so any leftover disagreement splits evenly between them.
- Those errors are usually not zero. They only reach zero if the distance between the two points in CAD matches the distance you actually measured on site.
Anchor two points when the layout has to line up with a direction, not just a spot — most often an existing wall.
Anchoring three or more points
Anchoring three points runs the best-fit analysis on just those three and drops every other scanned point out of the calculation. The result is identical to what you'd get by scanning only those three points and not anchoring anything.
So there is no reason to anchor three or more points. If you want the fit built on three specific points, record those three points. Anchoring them adds nothing, and any extra points you recorded are simply ignored — which is easy to misread as the app losing your work.
What anchoring cannot fix
Anchoring controls rotation and translation. That is all it controls. It cannot correct a mismatch between your CAD geometry and your field geometry.
If CP01 and CP04 are 40 feet apart in the file but 40 feet and 1 inch apart on the slab, no amount of anchoring closes that inch. The layout will align to the wall and then run long or short along it. Best-fit and anchoring can align to your control points; neither can tell you whether your control points were set in the right place to begin with.
Example: aligning layout to an existing wall
Your unit has an existing wall you have to build to. You measure two points along it — CP01 and CP04 — confirm they sit the same distance off the wall face, and scan both with the reflector. Then you anchor them.
What you get:
- The scanned baseline rotates and slides until it matches the real wall.
- Both wall points align with equal precision.
- Your other two points — interior layout marks — drift more than they would have without anchoring.
That is usually the right trade. A misaligned wall throws off every door, opening, and dimension downstream of it. Anchoring the wall points says: this wall must be correct, everything else can move around it.
The close-up shows the limit. In this example the control points were laid out with a tape measure, so the CAD distance between CP01 and CP04 didn't match the measured distance. The wall aligns — but the layout runs slightly long along the wall's length, and the leftover error shows up at the corners. Had the two distances matched, the printed layout would have matched the CAD layout exactly.
When to anchor and when not to
Anchor points are one of several Flexible Control options for jobs without usable surveyed control, alongside corner control, creating control at the reflector, and line stationing.
Surveyed control is always the preferred route. Flexible Control asks you to trust the accuracy of things that already exist on site — and walls are typically built to about ±1/8". Over long distances those tolerances stack up.
| Situation | What to do |
|---|---|
| Good surveyed control, errors within tolerance | Don't anchor. Let best-fit balance all points. |
| One critical location must be exact | Anchor that point. |
| Layout must align to an existing wall or line | Anchor two points along it — not one. |
| You're tempted to anchor three or more | Don't. Just record those points and station normally. |
| Control points are just bad | Fix the control. Anchoring hides error, it doesn't remove it. |
Never nudge the reflector off a marked point to make an error value look smaller — that hides bad control and makes the point unusable on your next station.
How this article simplifies the math
This is a deliberate oversimplification. The actual stationing solution is computed in one shot using a well-known technique for rigid-body alignment (the Kabsch algorithm, solved with a singular value decomposition), and anchoring is implemented by giving anchored points an extremely large weight in that same calculation rather than by literally locking them and rotating around them.
The step-by-step here — find the centers, slide them together, rotate — describes what the result looks like and why the tradeoffs behave the way they do. It is not a description of the code. If you need the real derivation, ask your Dusty contact.