1. Preamble
The past ten years have witnessed the development of various techniques for capturing point clouds at an enormous speed. At the same time, 3D-mapping has reshaped itself from being a niche-market technology for experts to user-friendly systems that opened the market to laymen. A consequence of this market growth is the decrease of required investments for hardware while current developments also show the emergence of hand-held 3D-cameras and miniaturized laser scanners which are embedded into off-the-shelf smart phones. This trend will force current hardware manufacturers to reshape themself into a software vendor and solution providers.
Even though the performance of currently available computers has drastically improved compared to the early days of 3D-data acquisition, processing the captured data remains the major bottleneck. This circumstance is caused by the increased speed of sensors for 3D-data capture which consequently also increased data volumes. Hence, efficient processing strategies and data flow are the key for economic prosperity.
Most hardware manufacturers have developed their own proprietary formats for storing and processing point cloud data. In addition to solely offering hardware many also develop and market their own software, e.g., for registration and georeferencing, which at least partially transforms them into a software vendor. Potential clients can choose an all-in-one package from the manufacturer that includes the manufacturers hard- and software. If the user follows a processing strategy that includes third-party software, usually he relies on open formats such as PLY, E57 or LAS that the hardware manufacturer provides for export with their capturing software.
But one can observe that end users today are choosing devices from different manufacturers and solving specific tasks by using various software products. The industry came to the E57 compromise for structured and unstructured point-clouds, and LAS and its compressed version LAZ for unstructured point-clouds. Unfortunately, due to the rapid increase of the project sizes, converting into these formats and import into different 3rd party software systems is costly and requires a lot of storage space (see Appendix). Even a potential extension to a compressed E57 format is not a straightforward way due to loss of hardware specific properties and information. In contrast, specialized proprietary formats can provide much faster read and write access as well as optimized for visualization. This conflict would be solved if SDKs for the vendor’s proprietary formats are accessible for everyone. In addition, this solved the major issue of long-term storage and future access to point cloud-data.
2. The Petition
With this petition all subscribers “driven by the inspiration of a free and fair market” ask kindly all hardware manufacturers and software vendors to either provide Software Development Kits (SDK), format descriptions or executable code that consequently allow to have efficient read and possible write ability of raw and preprocessed (filtered and combined) point cloud formats, and all ancillary data1, in the best case free of charge and easy to license.
3. Benefit for hardware manufacturers
The subscribers of this petition understand the hardware manufacturers and software vendors concern to allow competitors and open 3rd-party software access to their proprietary formats because more revenue will be generated from packages that include hard- and software or several software solutions. However, it is obvious that vendor software cannot provide all solutions demanded by the market and also lack the ability to combine data from different hardware. Also due to the transformation of hardware manufacturers to software vendors they also massively benefit from open format access to combine different point cloud sources. We firmly believe that free format access is the key to push market growth of point cloud based applications which will be beneficial for all hardware vendors and of course their clients. This argument can be justified by the fact that a more efficient data flow leads to more profit for the user - which in turn can lead to faster investments in new hard- and software. In addition, competition will lead to more solutions, inspire science and all market participants to improve their products and make this technology one of the key drivers to the digitization of the world.
4. Benefit for end users / service providers
More affordable hardware will lower the bar for even more participants in the 3D-mapping market – which is expected to continue its steady growth in the coming years. In order to survive as a service provider to this market a vendor can only a) remain attractive to potential clients by offering low prices, since the majority of competitors will deploy off-the-shelf solutions or b) develop tailored and specialised services to specific client’s needs as a unique selling point. In order to achieve the latter, a service provider may combine different sensor sources with 3D-data, e.g. for mapping moisture or sound characteristics of a building to captured point clouds. This also means that a single software vendor may not be able to implement all desired solutions that service providers’ demand. Hence, the two aforementioned solutions can only be profitable if the exchange and data flow between all considered sources and software solutions is as efficient and smooth as possible.
5. System integration and robotic platforms
We also will see the system integration of cheap scanning hardware with autonomous carrier platforms like cars and robots as the drivers of innovation in the coming years. This requires open access to the raw data for sensor fusion to push this market forward that all participants can benefit from this technological shift.
6. The “evil” twin(s)
Converting point clouds into other formats requires considerable computational efforts and typically occupy (at best) approximately the same disc space compared to the original data volume. This circumstance is somewhat disturbing due to the fact that the geometric and radiometric information that is represented by different point cloud formats is essentially identical. The unpleasant consequence of this process is a largely redundant “evil” twin of the original data. From an ecological viewpoint this practice has to be seen critical due to the fact that valuable resources in form of electrical energy and disk space are wasted while (for no gain of insight), the economical conclusion is that users need to sacrifice a considerable amount of their profit (see Appendix). Another unpleasant reality in practice are different interpretations as well as redundancies of meta information which may be interpretable by one software yet causes another to crash. Hence, a vicious cycle emerges since the user is forced to export large amounts of data again – due to some flaws in the magnitude of kilobytes. Also note that exporting different formats is a re-emerging issue in practice since users often concatenate different software.
7. Scientific horizons
Science has always been a catalyst for innovation – particularly in the world of 3D-mapping. The success and impact of scientific research heavily relies on free access to pre- or even unprocessed data. Current hot topics that would immediately benefit from unrestricted access to raw data are for instance error modelling of laser scanners, classification and segmentation of point clouds, which includes strategies such as artificial intelligence and machine learning, as well as the design and implementation of multi-sensor-systems where the advantages of individual sensors synergistically compensate for their drawbacks.
8. Survival of the fittest?
Every lifeform on this planet is part of a relentless competition against other species in the everlasting race of evolution. Closed file formats can be seen as unconquerable boundaries between species which consequently prevents different species, or approaches in a technical sense, to compete. As a consequence, many software solutions (provided by hardware manufacturers) are quite similar regarding their range of functions and algorithmic backbone. Open accessible formats create competition on one hand, which yields in better products, yet also establishes symbiosis since data from different sensors can be processed with the most suited software.
9. Appendix
In order to show much time is essentially wasted when converting point clouds, indoor projects with 100 laser scans each (~10 million points per scan on average without RGB-values) that were captured with different scanner brands were analysed. For this, proprietary formats were converted to E57 by using the original software of the corresponding manufacturer (applying comparable or, if possible, identical E57 export settings). Since the projects do not contain the exact same number of points, the export times and data volumes were normalised to 1 billion points for the sake of comparability. All computations were carried out on a local hard drive (SSD). Since export times heavily rely on the applied equipment relative speed measures were computed in relation to the fastest solution.
| Manufacturer | Data volume (Proprietary files | E57) [GB] |
Required export time [min] | Relative export time [%] |
|---|---|---|---|
| A | 1.1 | 19.2 | 8:11 | 100 |
| B | 3.1 | 20.2 | 15:09 | 185 |
| C | 3.1 | 14.0 | 17:56 | 219 |
| D | 5.8 | 19.0 | 43:19 | 530 |
| E | 6.9 | 18.9 | 48:07 | 588 |
| F | 6.1 | 14.6 | 53:01 | 648 |