In the previous article, we explained why a control network is the foundation of a reliable digital copy of an industrial plant and a prerequisite for long-term data consistency. However, this is only the first step. Equally important is how data is acquired in the field.

In 3D laser scanning practice, attention is still often focused on parameters that look good in technical specifications: maximum scanner range, very high resolution, or declared single-scan accuracy. Experience shows, however, that these parameters rarely determine the real usability of the data.

This article is based on the long-term experience of the 3Deling team and on observations gathered by Paweł Dudek, CEO of 3Deling, over nearly two decades of working with 3D laser scanning — from the first technology tests to large, complex industrial projects.

3d laser scanning field measurements

Early 3D Scanning Experiences – A Lesson in Humility

I remember the “tests” of our first scanner — it was 2007. We set a very high scanning resolution, because “it has to be dense” for the data to be good and for nothing to be missed. There was a slight surprise that a single scan would take around 30 minutes, but we waited for the result.

The scan finished, the data was transferred, then the point cloud was “processed” and opened in Pointools View (back then, it wasn’t Bentley Pointools yet). It took a while, but finally there it was — a very “heavy” scan. The data was visible at a very long distance. We could even see a chimney of a heating plant that no longer exists, located several hundred meters away. It was impressive.

This situation took place almost 20 years ago. At the time, each of us already had some experience with 3D laser scanning and we carried out such measurements on a regular basis. Looking from today’s perspective, however, it is clear how much we were still missing back then — especially when it comes to large-scale projects.

Today, our survey teams perform thousands of scans on a single site, all registered within one coordinate system, often under difficult conditions and time pressure. And in the end, only one thing really matters — that the client receives the best possible data.

Why We Scan Differently Today

In practice, the approach to scanning looks very different today. And it is not because we want to scan “fast and carelessly,” close the project and move on. Quite the opposite.

To obtain the most complete and usable geometric representation of an object, the key factor is the number of scans and their placement, not the maximum resolution or range of the scanner.

Scan Resolution – Why “Denser” Does Not Always Mean “Better”

Very dense scans are simply “heavy” datasets. They are harder to work with — both due to software limitations and hardware performance constraints.

That is why individual scans are often filtered and their resolution reduced. As a result, a unified point cloud can be five to six times lighter, while being much more convenient to use — without losing information that is actually relevant for design work.

Scanner Range – A Parameter Rarely Used to Its Full Extent

Most scanners we use have a range well above 100 meters — one of them even up to 600 meters. In practice, however, the data is usually used from much shorter distances:

• indoors: typically up to about 30 m,

• outdoors: typically up to about 50 m.

The full scanner range is rarely utilized and usually only in cases involving very tall structures with no safe physical access.

Completeness of the Geometric Representation – The Key Parameter

This is the most important data quality parameter — and at the same time one that can almost never be achieved 100%. There will always be so-called “shadows” or blind spots — areas with missing data.

However, these can be significantly reduced by performing a large number of scans from different positions, heights, and distances. With hindsight, it is clear that the number of scans is the key factor influencing the quality of the final geometric representation of an object.

Number of Scans and Real Design Work

We often support clients who are preparing for plant digitalization projects in drafting tender specifications. We then see that less experienced investors tend to focus primarily on parameters that look best “on paper”:

• range (the further, the better),

• resolution (the denser, the better),

• accuracy (ideally 1 mm).

We understand this — we used to think the same way ourselves. That is why we try to “demystify” these expectations and draw attention to what truly matters. And that parameter is the number of scans.

Where an object is well covered with scans, with many scan positions and a sensibly planned measurement geometry, subsequent modeling proceeds smoothly. The data is clear, there are no “holes,” elements can be interpreted unambiguously, and the model is created quickly — without guesswork.

In remote projects, for example in the Middle East, insufficient scan coverage becomes a serious issue very quickly. When data is sparse or scans are taken from unsuitable positions, modeling and design work based on point clouds turn into speculation. Information is missing, discontinuities appear, and it is unclear “what is what.” In extreme cases, such data is simply unusable.

Missing Data Means Real Costs

When data is incomplete, problems arise:

• returning to the site to perform additional scans,

• sending someone with a camera to take manual reference photos,

• accepting simplifications and uncertainties in the model.

Each of these options means additional time, cost, and risk of errors.

That is why, in practice, instead of performing a small number of very dense scans, we focus on a large number of scans with slightly lower resolution but good object coverage. This allows us to:

• obtain complete geometric data,

• minimize blind spots,

• create good conditions for modeling and design work,

• significantly reduce the time needed to interpret the data — designers do not have to guess what is where, because everything is already clear at the point cloud stage.

Unified Point Cloud and Working with Data

All scans are combined into a single unified point cloud, usually additionally filtered (e.g. to 5 mm). This unified cloud is used for 3D modeling and further design work.

At the same time, all individual scans with color information and panoramas are preserved and can be accessed at any time — for example via WebPano. This is a major advantage, especially for complex installations, where checking details, heights, and spatial relationships is crucial during design.

What to Look for in a Request for Proposal?

When selecting a 3D scanning provider, it is worth looking beyond hardware specifications.

Not only at:

• resolution,

• range,

• manufacturer-declared scanner accuracy.

But above all at:

• the estimated number of scans for the object.

This is one of the best indicators of the real quality of the data you will receive. A higher number of well-planned scans means fewer uncertainties, faster design work, and real savings in time and cost throughout the entire project lifecycle.

Summary

Resolution and scanner range are important, but they do not determine project success.
The number of scans and their placement have the greatest impact on the quality and practical usability of the final data.

Other important factors include the accuracy of the unified point cloud and a properly defined coordinate system — topics we will cover in the next articles of this series.

Planning 3D laser scanning or industrial plant digitalization?

If you want your data to be complete, consistent, and truly usable for design and engineering, the scanning strategy should be defined before any fieldwork begins.

At 3Deling, we help clients plan the number and placement of scans so that data quality translates into real time and cost savings throughout the project lifecycle.

Contact us to discuss your facility and project requirements.

The post Data Quality in 3D Scanning: Why the Number of Scans Matters More Than Resolution appeared first on 3Deling - Experts in 3D Laser Scanning and Point Cloud Processing.