A high-level multidisciplinary team to change the game on 3D point clouds.
Founder of TagLabs, after 10 years of experience in HVAC and industrial refrigeration, and 8 years as a 3D laser scanning service provider (founder of Liber-D in 2011).
Engineer from the Ecole Centrale de Nantes, Robin is responsible - among other things - for the development of the ScanSap 3D Vulkan engine.
ESIEA engineer, Aurélien joined the team to reinforce the 3D development, algorithms part and intervenes on multidisciplinary aspects (server, database ...).
A graduate of the Ecole Normale Supérieure de Lyon, Lucas works on mathematical aspects, and in particular on object recognition.
Each functionality is examined in detail in order to detect the slightest defect, whether on the expected result or its ergonomics.
We strive to offer new features. If a function basically repeats what exists, that does not interest us: it must at least do much better!
We favor substantive work over big declarations and cosmetics.
We master what we produce and are ready to perform live demonstrations on your point clouds.
With the rise of the power of computers, one is quickly tempted to produce a medium code which will be compensated by the computing power. We have chosen to optimize the code in order to make the most of medium configurations, thus fighting against planned obsolescence. We are also opposed to online point cloud streaming solutions, which are the opposite of digital sobriety.
3D surveys have been around for more than 2 decades. First reserved for advanced sectors or for research (archeology, heritage, oil & gas, nuclear), they have become more democratic in recent years, boosted by the arrival of BIM.
The point clouds generated are fascinating and at the same time frustrating: fascinating because they contain millions of measurement points, frustrating because their volume (gigabytes) makes them difficult to exploit by those who carry out technical studies.
So much so that they often return to the field with their measuring tape while a 3D survey has been done.
Software publishers, engineers, and researchers, have been working for several years on solutions that simplify the handling of these data and their transcription into a digital model.
The different current approaches are:
Convert the point cloud into a mesh:
The point clouds are decimated to lighten them, then triangular surfaces are created and textured. In this way, it is possible to divide the weight of the data by a factor of several tens, while having a very good visual quality. Unfortunately, this is done at the expense of precision, in particular on elements that contain edges, which is common in industry (structures, equipment, flanges, etc.). Choosing a building mesh means closing the doors to the accuracy required by all technical contractors.
Attempt to create the 3D model automatically from the point cloud:
Some are working on an automatic classification of point clouds in order to distinguish walls, floors, stairs, furniture etc... then to model them through families of objects. If this approach can lead to uses in classic BIM, it is very insufficient in complex environments. Indeed, recognizing a table is one thing, recognizing a piping and its standard is another. The semantics required for technical items is much more extensive than that necessary to recognize an object by its shape. This is understandable, since no laser scanner can distinguish 2 carbon steel pipes of different standards. Only the person in charge of the project, through his knowledge and the analysis of other elements (P&ID, technical documentation), is able to make a relevant choice. The automatic modeling approach also has the disadvantage of lower accuracy, making the model incompatible with the requirements of companies that do offsite prefabrication.
Our approach is different:
We think of all stakeholders, not just those who intervene in the pre-project phase.
For us, the point cloud is essential at all stages of operations. Instead of considering the simplification of the cloud as the only solution, we opt for a hybrid mode in order to:
Make the point cloud - even large - usable on a standard machine and use it to the best of its accuracy in order to meet the needs of the most demanding companies (pipefitters, process equipment, structures).
Accelerate the modeling of the items essential to the project by combining automatic detection algorithms and the addition of technical information by the user. It is therefore a supervised modeling.