Both methods of 3D documentation ultimately produce the same output, a point cloud, but each generates that output in a fundamentally different way. This difference leads to common questions such as, is one method better than another, or which technique should you use in a specific situation? I aim to put you through a quick primer on both approaches, to explain how they work, the benefits, the limitations and choosing the right 3D modelling solution for the job. My hope is after reading this, you will have learned the basics to make informed decisions on when to use each technique.

WHAT ARE SOME OF THE USES OF 3D MODELLING FOR HERITAGE?

Excavation site plans and documentation
Remote site interpretation
Archival recording
Conservation
Artefact recording, reproduction and re-creation (if damaged)
Damaged site re-construction or preservation
Monitoring of at-risk sites (makes for an easy to interpret tool to prove the level of damage that is occurring)
Public display and marketing materials
Tourism and public interpretation
Landscape-scale interpretation
Design tools for architects
Re-creating historical landscapes, artefacts and sites
3D site experiences in Virtual Reality and Augmented Reality
3D printing artefacts and even whole sites for display.

And many more; the list is growing with people’s imaginations and increased understanding.

What is a point cloud?

A point cloud is a three-dimensional collection of positional data. It is the visual representation of the many points a 3D optical measurement system has collected. These point clouds are made up of millions of points, each having a coordinate in Cartesian space (x, y and z). Below is an image of a point cloud; as you can see, this is not a continuous surface but a collection of millions of points that make up the basic shape of the object captured.

Point cloud view of the eastern arch at Avon Dam, New South Wales. Notice how you can see through the cloud to the other side

This collection of points can be joined together to make a continuous surface using a process called "meshing". This will allow the user to have a watertight model which approximates the captured surface accurately.

How laser scanning works

Laser scanners (LS) or LiDAR use an active sensor to collect data. Active means that it emits a laser pulse and measures the time it takes for the pulse to return to the scanner after bouncing off a hard surface. As we know the speed of light very accurately, the LS multiplies the time taken for the pulse to return by the speed of light and then divides that number by two (for the travel distance to the object and the return). The result is an accurate distance measured in metres. The LS also knows the horizontal bearing it is pointing to and its vertical angle with a horizontal plane; therefore, using simple trigonometry, it can compute a cartesian coordinate for that point. The laser spins vertically while the actual body of the LS turns horizontally, recording millions of measurements every minute. The video below illustrates the process (Source: Faro)

How Photogrammetry Works

Photogrammetry, on the other hand, is a passive sensing method, meaning it does not emit any energy to generate a 3D model. Instead, photogrammetry uses complex computer algorithms to detect and extract similar pixels from overlapping images. This technique is called stereoscopic imaging and is very similar to how having two eyes in different positions allows you to see depth by comparing the differences in those images.

Thus the primary technique involved in photogrammetry is the collection of hundreds of overlapping images. This collection requires a high-quality camera and decent storage solutions. However, whilst easy to generate, the images do not have an inherent measurement system like LiDAR. As such, the model has to be georeferenced or scaled using multiple known points and accurate measurements captured in the images.

The creation of the photogrammetry model then relies on a complex computer program to align the images and extract similar points from multiple photographs. However, due to the large image sizes and nature of the processing, this can be very time-consuming and hinges on the computer's power and the operator's experience. The image shown in "what is a point cloud" is an excellent example of the output of a photogrammetry model before it has been meshed and textured.

Benefits and limitations of both methods

Laser scanning

Benefits

High accuracy
Inherent scale - no additional surveying or measuring is needed unless georeferencing is required
Efficient when compared to detail captured - i.e. vs a land survey with a total station
Captures a large area around the scanner
Can penetrate through vegetation
People and machinery are easy to remove from scans
Lighting conditions are not an issue
Requires little processing time - i.e. almost ready to go when back in the office
Can deal with buildings very well, along with things that have slight movements like trees
Large scenes very easy to document
Indoor locations are simple to construct

Limitations

Very expensive
Requires specialised equipment
Can be large and difficult to carry into hard-to-access areas
Does not have a high visual fidelity
Requires training and planning to operate
Can be tedious to undertake
Requires careful placement of the laser scanner to ensure sufficient overlap between scans and no missed areas

Photogrammetry

Benefits

Very cheap to undertake
High accuracy - if scaled correctly
High visual quality - with good conditions and photography, models are very close to real-life representations
Can use equipment already on site (smartphones and cameras) - model quality will decline with inferior equipment and technique
Can be easily performed with a drone
Can be undertaken very quickly over small areas and objects
Simple to understand

Limitations

No inherent scale; the model must be scaled and georeferenced using specialist tools
Highly impacted by shadows and lighting conditions
Cannot be performed at night or in dark places without additional equipment and specialist techniques
Requires constant vigilance to ensure all areas are adequately photographed - can be very boring
Large scenes are time-consuming and challenging to undertake
Processing is time-consuming
Most models will require editing to remove spurious areas
Struggles with glass, vegetation, corners and large single colour surfaces
The quality of the model is unknown until it has been processed
You may have to perform a great deal of human correction during the processing phase if photographs are of sub-par quality - This usually lowers the accuracy of the model

Textured photogrammetry model of the eastern arch at Avon Dam, New South Wales

Questions to ask yourself, before committing to an option

Is accuracy more important than visual quality?
Is the surrounding landscape required to contextualise the object?
Are there highly variable lighting conditions present on the site? Have to perform the work at night or in a dark cave? Is the worksite open and lacks shade?
Is there a great deal of vegetation present around the object?
Is there a lot of work that must continue during your survey around the object?
Is the object likely to move slightly?
Are you trying to document the interior and exterior of a building?
Does your budget cover the rental of expensive laser scanning equipment?

If you answered yes to most of these, laser scanning is likely to be the best option

Is the object difficult to get to? Will it be difficult or dangerous to carry expensive equipment onto the site?

Is visual fidelity the most important outcome?
Is the survey area likely to be free of people and machinery?
Do you have experienced photographers on site?
Do you have time to process and edit the resulting model?
Are the lighting conditions good? If not can the object be shaded?
Do you have a smaller budget for documentation?

If you answered yes to most of these, photogrammetry is likely to be the best option

FINAL THOUGHTS

Both methods have their benefits and limitations.

If your budget allows, the optimum solution is usually to use both in co-operation. With a laser scan forming the metric basis for the model and a photogrammetry texture being placed on top for visual fidelity.

With this combination of techniques, you obtain a high-quality output in both accuracy and visual quality.

Below is an interactive 3D space containing a highly accurate photogrammetry model. As can be seen, it is possible to get similar results to laser scanning. However, this does require experienced operators, specialist equipment and good conditions. For most archaeological uses, tolerances are not as small as in the engineering sector, so photogrammetry will usually suffice. Although, proper technique must be applied, or the resulting model will be of sub-par quality.

I hope you managed to get some information from this blog, if you have any questions or something is still unclear - please get in touch, I’d love to have a chat. You can contact me at will@heritagespatial.com.au or through the “get in touch” button at the top of the page.

Should you use Laser Scanning (LiDAR) or Photogrammetry for your Archaeology Project?