CAD: Point Cloud technology
Using point cloud readings, it is possible to include 3D-scans in CAD models
Building surveys are problematic. Even with the assistance of sonic measuring devices, they can be difficult, sometimes treacherous, with complicated access arrangements and safety procedures, while the process of drawing up surveys is often regarded as a tedious chore.
No wonder architects usually identify surveys as additional services
Disagreements arising from measured surveys cause endless recriminations and, according to John Wevill, head of non-contentious construction at Fishburns Solicitors, they are a major source of litigation involving architects, second only to design liability. No wonder architects usually identify surveys as additional services in their contracts of appointment, often leaving this to specialists. Wouldn’t it be better if this whole routine could be automated?
The information in the image above of London’s Kensington Library was gathered by taking a 3D-scan, which recorded billions of points on the building’s surface to produce a data file, known as a point cloud. This scan also recorded reflective intensity and RGB (red, green and blue) values. Hence the strange colours.
According to architecture graduate Faraz Ravi, founder and director of UK company Pointools Ltd, which has developed software to manipulate and organise unwieldy point cloud data, ‘the first production laser scanner for architectural scale scanning was released in 1998, although it was mainly used in petrochemical industries’. I downloaded a trial version of Pointools View software and discussed the use of point cloud in surveying, design and presentation with Ravi.
Take a stroll through the Pointools model of Kenilworth House in the video below.
Architects are unlikely to be directly involved in 3D-scanning, Faraz Ravi explains, as this involves training and costly equipment. Paul Burman, director of Hester Architects, which uses Pointools, also notes that tripods are required and that if scans are taken from several positions, the assembled data will be more satisfactory. This will minimise the crumbling effect seen in the image of Kensington Library.
Architects are more likely to be involved at the next stage, when the point cloud data is loaded into a CAD environment. But there is disagreement about the usefulness of this data. Architects are accustomed to working in CAD files, which comprise vector geometry (lines and raster data (2D configurations of points). Although a point cloud can co-exist with both types of information, in a CAD environment such as Pointools, it remains fundamentally different.
Point clouds can be manipulated, but they are less amenable to typical vector geometry operations, such as slicing, trimming and stretching. Ravi argues the case for leaving survey data intact and recommends that only its visibility should be adjusted. One benefit of a Pointools file is that it retains key information such as the type of scanner, the original scan file name and the scan’s date and location.
There is also a traditional view that the principal output of measured surveys should be orthogonal, dimensioned drawings. Tim Leach, a BDP director based in the London office, says they do not use 3D survey data in the office. I can understand why. It can be immensely frustrating to develop a design with 3D survey data alone.
Two-dimensional drawings, often incorrectly dismissed as ‘dumbeddown’ by 3D CAD enthusiasts, are actually sophisticated analytical tools, which enable architects to isolate design elements in order to assess and compare their sizes and proportions. These properties are difficult to judge in 3D-views because of the distorting effects of parallax and foreshortening.
Design and Presentation
Since point clouds occupy a parallel universe to vector geometry and raster data there are limitations on their use in design and presentation. ‘Converting a point cloud to a mesh is a non-trivial problem. For this you need specialist software and these packages cost around £10,000. Once you have a mesh it will probably be too dense to load into 3ds Max or other similar software,’ says Faraz Ravi.
At the design development stage, it makes sense to model proposals in vector geometry and use point clouds as a reference. For example, Hester Architects has used Graphisoft’s ArchiCAD to model proposals for its Kings Point mixed development in Reading and this model is imported into a Pointools file, which then uses point clouds to represent existing buildings. This enables the architect to check alignment of the existing and proposed building envelope and evaluate it in context.
Paul Burman explains that the dots of the point cloud can be enlarged to produce a more solid appearance and that the RGB colours can be interpolated to produce a naturalistic representation of existing finishes. Nevertheless, the architect chose not to include point clouds in the final drawings.
It is possible to get visually homogeneous results using image-editing software and dense point configurations can look identical to lines. The alternative is to express the ethereal qualities of point clouds in images which resemble air-brush rendering.
Pointools is approachable and relatively inexpensive. It can import dwg, dxf and 3ds files as well as produce animations. Nevertheless, it is niche software which is only used by a few UK architectural practices. Pointools 4 Rhino can produce accomplished renderings: TMJ Interiors’ image of a restored ceiling (left) is a good example.
Last month, Bentley Systems announced that it had licensed Pointools’ powerful Vortex API engine. Subscribers to Bentley Systems’ SELECT Microstation programme will be able to use this from 2010 to reference and coordinate point clouds in Microstation files.
Faraz Ravi notes that this update will include point cloud manipulation tools. He suggests that in the future, point clouds may be able to cast shadows and be used in clash detection analysis. Bentley Systems says that it has yet to discover the full potential for point cloud models within its portfolio of applications.
If only someone could devise an economical way of converting 3D scan data into vector geometry, life would be so much simpler. As things stand, point cloud technology remains a source of extraordinary images which provides limited assistance to the processes of surveying and architectural design.