An introduction to and overview of 3D Scanning
Scannerbox is primarily intended as a resource for interested parties from the Cultural Heritage and Creative sectors who wish to employ 3D scanning in their activities, whether for preservation or creativity support.
3D scanning is not a new technology by any means – it has been employed by the several fields and sectors for several decades. Archaeologists have used 3D scanning techniques to capture artefacts and monuments, which has enabled them to study and preserve them in ways that would have been otherwise impossible. Similarly the engineering sector, guided by the field of metrology, uses 3D scanning for accurate measurements, both for production and reverse engineering.
The creative industries have certainly not lagged behind: 3D scanning is commonly used to create 3D virtual assets of environments, objects and people for use in film and digital games. And Museums have embraced the technology when it has been placed within their reach.
However, while 3D Scanning has seen wide adoption, it is still a technology that is normally out of reach of most people and organisations. There are several reasons for this: the primary being cost, closely followed by the complexity and diversity of the required skill sets.
The cost of 3D scanning is directly related to the technology involved. Until relatively recently, 3D scanning technologies have relied on range-sensing laser technologies. Laser-based 3D scanning, while tremendously accurate and reliable under ideal circumstances, is costly and has several disadvantages. The equipment itself, while readily available on the commercial market even in handheld form, is extremely costly, and relies on proprietary hardware and software that require specific training, have narrow applications, and questionable longevity. Furthermore, the laser-based technology is unsuitable for several types of objects and is unable to scan shiny and translucent surfaces and objects. And the effects are further compounded when the scanner is used in uncontrolled environments, such as outdoors.
The same issues arise with another popular technique: Structured Light Scanning (SL). This technique involves projecting a known pattern (most often a series of black strips similar to a barcode) onto an object using a lighsource such as a projector. Assuming the object is not flat the projected pattern is deformed, similarly to how a a tablecloth or sheet may deform when covering an object. Image you have a tablecloth with a picture on it that you are familiar with, if you were to drape that sheet over a chair, the sheet would fold and curl around the chair and the picture would similarly deform. As you know what the picture is supposed to look like, you can visually understand what the shape of the object it is covering looks like. This the fundamental technique of SLS scanning. When the image is projected onto an object, one or more cameras in a stereoscopic arrangement measure its deformation and – just like the human eyes and brain achieve depth perception – derive information about the object and can reconstruct its shape.
While potentially more affordable compared to laser scanning, the structured light method still relies on relatively specialised setups. While SL scanning setups can be created with off the shelf cameras and projectors and free software, there is still considerable effort involved in creating the project/camera assembly and achieving the necessary calibration. The results of SL scanners are on par with laser scanners for a fraction of the cost but share many of the same drawbacks, such as a considerable hardware investment and an inability to tackle certain types of objects such as shiny or translucent surfaces and objects with deeper cavities where the laser or projected light cant reach.
From the above, it can be said that the first step in any 3D scanning approach is to find a way to ‘measure’ ranges, and thus derive depth information. Laser scanning achieves this by bouncing laser beams of the surface of the objects. Structured Light scanning accomplishes it by measuring the distortion of a known pattern over the object. In both cases the scanning devices are measuring the ‘depth’ of the images they ‘see’. From this depth they can derive the geometry of the objects and environments that they are ‘seeing’.
Both techniques are very effective, but rapid advances in Computer Vision technology have introduced new ways of measuring – and understanding.
Photogrammetry – A Flexible and Affordable 3D Scanning Method
Rapid advances in computer vision and image analysis technology have introduced new ways of measuring depth from images. This has enabled techniques such as photogrammetry, that used to be a labour-intensive manual process, to be both automated, and applicable as massive scales. Photogrammetry is, as the name would imply, the science of taking measurements from photographs. Just like the techniques described previously, having one or more images of an object from different angles can allow you to derive the geometry of that object. Modern computer vision techniques can help you do so with massively increased speed and accuracy.
All that is necessary are several good images of the object or environment that you wish to scan. The source of this images can be any camera, whether an inexpensive web camera, top-of-the-line SLR camera, or your mobile phone’s camera. Commonly available photogrammetry software handles the task of analysing the images, comparing them for similarities, and using them to measure the dimensions and geometry of the subject and generate a usable 3D model. The results can be just as accurate as the aforementioned techniques, and are often better due to the use of the images to generate photorealistic textures for the models.
Scannerbox focuses on the use of Photogrammetry technique, and covers in detail all of its currently relevant aspects, including the skills involved, the equipment and software, and the practicalities of its use.