The 3D design and engineering world is bringing together solids/surfaces with meshes, lattices and point-based data like never before. How will data translation technology adapt to bring together these often disparate and incompatible geometries and make them interchangeable?
John McCullough, marketing general manager, Kubotek
With the addition of new 3D data types in the product definition arena, it is more important than ever for design and manufacturing teams to clearly define and communicate the authority dataset.
Unless there is a specific need to limit accuracy, that dataset should always be the most precise and complete type of geometry available, usually a solid.
Many 3D technologies, from finite element analysis to additive manufacturing, operate on mesh models that can be derived from the authority solid using widely available and easy-to-use software.
Any mesh, lattice, or point model derived from a solid only approximates the original. Some of the precision is lost.
Depending on the intended process, that loss can be absolutely fine, but it’s best to leave that decision up to the process expert who requires the derivative. A related problem we have seen is CAD users who are not aware that STL files are not just a translation to a different CAD format.
The precision lost when a solid CAD model is converted to an approximate STL mesh can be significant and a supplier receiving an STL can’t put that precision back.
It is best practice with all translations and conversions of critical 3D data to include independent model comparison software in the process, in order to validate that the derivative has not lost any data and is within required tolerance of the original.
This practice is widely used by aerospace suppliers working on aircraft programmes that use model-based definition (MBD) as the authority data set.
There are evolving technologies to convert mesh, lattice and point data into solids, but it is important to understand that these are not meant to be used in back-and-forth translation.
The proper use of them is when a solid is needed and the original data was in mesh, lattice or point format – from a scan of a physical part or from animation artwork, for example.
Roman Lygin, CEO & Co founder, CAD Exchanger
Precise B-reps and polygonal meshes often go hand in hand.
This drives the necessity for consistent data structures and algorithms for both, including support of multiple part representations and multi-LODs (for example, availability in JT or STEP formats), simplification and decimation, as well as particular geometry recognition (such as NURBS-to-analytical surfaces) or cross-references (for example, per-face triangulations or mesh-to-face indexing).
Both require parallel algorithms with good scalability to support growing geometry complexity and assembly sizes.
Still, given very different sources of origin (such as design versus scan), the representations can hardly be called really interchangeable or equivalent. Much richer information encapsulated in B-rep often makes it a better master representation, which provides the required accuracy of a mesh.
Mesh-to-B-rep conversions are certainly in demand, but their implementation will take time to achieve quality and speed comparable with B-rep-to-mesh.
Unlike professional CAD developers, domain-specific software vendors and end users often do not have a deep understanding of the peculiarities of 3D formats or representations.
It is not uncommon to hear this: “I need a STEP file from this STL one.” So it often takes extra effort to educate users on which formats to use for more efficient data exchange.
Format standardisation committees can contribute by encouraging support for dual (or multiple) representations. Hardware vendors, meanwhile, should be taking an interest in leveraging more powerful 3D formats and not just those that are easy to use.
Overall, it will take a concerted effort from multiple participants in the workflow, not just CAD conversion technology providers, to achieve the greatest impact on efficiency improvement.
Kentaro Fukuta, general manager, global business, Elysium
Compatibility of raw data between solids/ surfaces, polygon/ mesh, and point clouds is a minimum requirement for data translation, but it will not be enough to fully leverage the value of data.
It becomes more important to include the context, so as to fully equip the consumer with the deliverables of an application in which the data is being applied.
For example, in order to be able to apply manufacturability checks, it is important to recognise not just the features but also the design intent, which may mean having an understanding of the data beyond ‘raw’ geometry.
It is also important to be able to recognise specific areas of relevance in situations when data may be overly abundant.
For example, in a scanned point cloud environment, the ability to extract the objects targeted for use in a specific context is necessary to make efficient use of the information.
Interoperability, with the goal of adding value to the raw data, will be more essential when we accelerate MBD/MBE to establish robust digital end-toend processes.
When more and more product information is digitalised with added value, such as complete manufacturing instructions, data will be better connected, automated and consumed across the organisation and the supply chain.
Interoperability will be more platform-centric in the near future; the mobility and modularity of interoperability software will become significant, and can then serve to fully utilise the variety and types of 3D digital data and tools in use.
We should not be tied to specific formats/systems or tools going forward. It will be a mission for interoperability software to remove the barriers between process, companies and organisations.
Ryan Dugmore, head of consultancy, Theorem Solutions
As CAD has evolved, so has data translation. At one time, it was all about lines, points and surfaces, and then later, product structure and solid bodies.
With early virtual reality (VR), and when the PLM vendors offered desktop viewers, it meant polygon data had to be created and consumed.
Today, mixed-mode data is handled, and bodies made of surfaces, B-rep solids and polygon meshes can be created. If the target system supports those forms, it can be consumed. Scanned data can be created and meshes produced.
It’s now feasible to read and create an assembly composed of parts (bodies) made up of surfaces, solids and polygon meshes. The issue is what will the target CAD system do with it? If the target CAD can define a mixed assembly, then it is feasible to read and write that data.
How well the modeller will deal with a ‘mixed’ assembly is a question for CAD vendors.
Another use case is support for data in extended reality (XR) applications, which, being based on gaming engines, means polygon meshes. That data can be created today, and if needed, optimised to improve the XR experience. This begs the question again: If a user modifies the source CAD data in XR and outputs it, what would the CAD system do with it?
The answer for systems that support hybrid modelling is they could consume and model with it. So when use cases demand, data translation will be the enabler for sharing mixed-mode data.
Peter Kerwin, Parasolid product manager, Siemens DISW
The rise of mesh based technologies is challenging software vendors to mix and match mesh data with long-established boundary representation (B-rep) data structures and algorithms in complex product design workflows.
For data exchange vendors and standards organisations, the shifting task is to understand how each vendor is changing their data structures, in order to identify commonality for developing standards and to understand where fundamentally different approaches require data translation.
Parasolid’s answer to managing mesh data was convergent modelling, which integrates meshes into B-rep data structures and algorithms as a new surface type with topology, delivering new capabilities without disruption to existing functionality or legacy data integrity.
We’re also delivering ball-and-rod lattice support employing the same evolutionary approach. Our commitment to data openness was central in the design of both.
For example, with a single, unified data structure, it’s straightforward to update the XT file format to enable the Parasolid community to exchange new data formats in XT files.
Siemens Digital Industries Software doesn’t see other major vendors following this unified data structure strategy, so we expect end users will see new data formats causing potential interoperability headaches.
To support interaction with non-Parasolid licensees, we publish the XT file format and data exchange vendors have access to functionality for building the Parasolid end of mesh-enabled, application-toapplication translators.
For application developers, the Parasolid/STEP translator toolkit uses AP242 definitions to export/import models as collections of NURBS surfaces and mesh surfaces with connectivity data to rebuild the model on receipt.
For import, Parasolid reads STL mesh files and has extensive analysis and repair functionality to improve mesh quality where necessary.
Mark Gammon, technical director and CADfix product manager ITI, a Wipro company
The 3D digital geometry landscape has changed rapidly in recent years, with novel forms of geometry reaching industrial maturity.
High fidelity 3D scans are now commonplace. Super-flexible subdivision surface geometry powers intuitive free-form design tools. Lattice-based implicit surface geometry driven by simulation algorithms create designs unimaginable with traditional MCAD tools.
This rapid expansion of the industrial geometry frontier is exciting and promising, but geometry exchange weaknesses are limiting opportunities.
The existing MCAD-based toolchain supports a vast ecosystem of digital geometry consumers. To tap into these, robust bidirectional MCAD connections need to be developed.
However, fundamental differences in the maths supporting the 3D geometry representations make this a nontrivial technical challenge.
Access to the mathematical definitions of the various representations is an essential pre-requisite. STEP (mostly) provides this for traditional MCAD, but the other forms are either too immature or fragmented (for example STL, AMF, 3MF and so on).
Continuing innovation will outpace the standardisation process, which leaves open/ royalty-free programmatic access as the near-term solution.
Pixar provides an OpenSubdiv toolkit, Autodesk provides an FBX toolkit. Other providers of novel geometry need to see open access as an essential part of their wider industrial adoption strategy through the collaborative R&D it enables.
A hybrid geometry engine capable of simultaneously holding a design in multiple linked representations is the approach ITI is taking to upgrade its CADfix data exchange product for the multi-representational future.
Standards, open access and collaborative R&D will be essential enablers in this and the innovations of the 3D digital geometry industry in general.
