Is anyone here thinking about or working with IFC roundtrip in any way, or is this a “no-go-zone” for buildingSMART?
I see there is a change in the semantics of the scope of IFC4 DTV from excluding “round-trip scenarios” to excluding “data-roundtrip” - between the old site and the new. Is there a difference where there is a scope taking care of “geometry-roundtrip”?
As I see it, we need to talk about this issue, before our native models collapse and we only have IFC to lean on. I have experienced several “near death” situations, and I feel this issue is starting to come really close to a campfire based on petrol.
I would really like to get in touch if there are some open-minded souls having the same thoughts and feelings as me on this topic.
Jotne has for more than 30 years been working for customers to deliver information ownership based on open standards such as ISO 10303 STEP and ISO 16739 IFC.
Information ownership without roundtrip is not possible! Jotne therefore facilitates the management of everything in todays and future IFC, and we need applications to import export data without information loss. We therefore support this work Bjørnar.
Why is buildingSMART investing in model development for decades if it only was for geometry? This is already more advanced available in STEP since 1994 and well deployed in the manufacturing industry?
Agency for Public Management and eGovernment (DIFI) in Norway is mandating authoritative databases based on open standards. This means IFC and that data must be imported and exported in full between processes and applications.
The National Archives of Norway (Arkivverket) has allowed IFC as a legal archive format. From 2018 it is possible to archive models instead of PDFs or similar. Long term archiving and retention is impossible without roundtripping capabilities. Here we are supporting the ISO 14721 Open archival information system (OAIS).
Speaking as a volunteer on FreeCAD, perhaps we can be more transparent on what we support as roundtrip as a vendor. As users are more aware of what is advertised by vendors, they will start to demand more, perhaps.
… if there was some sort of autogenerated suite of unit tests from the documentation perhaps …
I’ve also been interested, for quite some time, in realizing some type of roundtrip workflow via IFC.
@yorik and I have been exploring a simple extrusion and material exchange between FreeCAD and Revit, to create small scale details like these and these.
To tease out what we can and cannot do, we log our test files here–a fledgling MVD, so to speak, we call FreeMVD.
For us, since these projects only tap into a small, compartmentalized aspect of a project’s overall documentation, it allows us to more easily realize the benefits of rountripping. That is, verses trying, right out the gate, to roundtrip the entire model and all its bells and whistles.
We feel if we can start a workflow ‘wedge’, and demonstrate it’s advantages, we can widen that wedge over time, adding more functionality, as we go.
Also, in this way, we are developing an evolving MVD that is directly born from a working workflow, vereses, an MVD developed in apriori fashion.
What we’ve been exploring so far with @Theoryshaw is basically reduce the scope of what we’re exporting and importing, which raises much the level of control that we can have. For example we’ve been trying things like “all objects must be extrusions” or “all objects must be IfcBuildingElementProxies” (so some BIM apps don’t add fancy unwanted transforms) or “no building hierarchy in the file” (“flat” file, only contains objects, no storeys or anything). These give amazingly good results, suddenly you find yourself with very trustable IFC files where you can spot any error very quickly.
There are things pretty complex to get right, specially given the particularities of each application, but sticking with simple stuff can work well. Of course it also depends on the level of “accuracy” you want/need. Some things inevitably change between one export and the other (order of vertices in the file, #id number of each entity, etc…) so it’s pretty hard (impossible?) to get a file 100% identical after a new export/import cycle. But the important stuff (objects, geometry, properties) can be reliably preserved I think
I know this is an old issue, but worth providing details on how this has evolved since 2019.
Since then, I have come across numerous projects who have wanted to natively author IFCs - effectively round-tripping with 100% data transfer and no data or geometric loss.
The BlenderBIM Add-on is one of the few options which now do this - where an entire project is 100% in IFC and round-tripped, with no other formats, data mappings, or otherwise required. It is not the only application that does this, but I believe it is the most “comprehensive” (i.e. everything from structural, cost, 4D, to parametric solid manipulation). A developer API within IfcOpenShell is also provided for those who want to build their own toolkit or pipelines which work with native IFC. Another proprietary example is the usBIM+ Viewer, with some restrictions on what it is capable of editing.
It is an exciting time, and relatively uncharted territory. Feel free to get in touch @Nordlenningen - I’d love to demonstrate what’s possible and where it’s headed.
I completely agree. This is already happening and is causing great concern where I work. “Read-only” IFCs treated like PDFs, or multiple snapshots limited to an MVD just don’t quite cut it anymore.