Lost in CAD Translation

LostInCADTranslationThe amount of choice in the ever-expanding CAD universe is not without a strong caveat: the persistent interoperability woes for which no one authoring solution can fully conquer. CAD interoperability is treated as an incurable disease, the unfortunate cost of doing business within a diverse supply chain of industry conglomerates and nimble design shops adrift in an ocean of variable formats and constantly flowing technology. Coping mechanisms are in abundant supply, be it a variety of neutral formats, direct modeling tools, direct and indirect translation, or the good old brute force method: old-school remodeling. But the pain remains. Reliable direct translation seems to be the preferable solution, especially with regards to preserving design intent, but things are not quite so simple. When it comes to properly handling the abundance of CAD file formats in the wild, engineers and companies too often find themselves staring longingly through a window across a foreign landscape, knowing they are hopelessly lost in CAD translation.

When it comes to direct translation, traversing geometry kernels cleanly is not a trivial exercise. Translation software is necessarily complex, relatively expensive, and unfortunately creates imperfect results. It’s hard work battling asymmetrical math models, waging wars of attrition with vertices, B-reps, and NURBS. You’re outgunned right from the start, with nothing but a 64bit floating-point shotgun. Those infinitesimally small rounding errors are ganging up on you; it’s quite literally death by a thousand tiny decimals. Recognizing the challenge, the best translation software tools include a variety of healing methodologies to assess and repair the damage incurred when moving across topologies. But the expense and difficulty of properly leveraging these tools may be a bridge too far, especially if you don’t have the time for simultaneous PhD’s in Computer Science and Math. So this is often a job for specialists. Or you could just mash the translate button, toss it over the fence and hope for the best.

Could things get worse? Yeah, they’re worse. Larger companies purposefully limit their interoperability exposure through format mandates to their supply chain. It’s generally a sound data quality strategy, if you can exert that level of control on your subcontractors. But that’s where interoperability woes place a uneven burden on the smaller firms. What happens if you are deep in that supply chain? As you move further down the chain, the companies get smaller, but the potential of having to deal with multiple foreign CAD formats from varied contractor requirements increases. It’s all due to less leverage over other players. Such companies can find themselves in a difficult situation, unable to both afford and effectively utilize the available repertoire of translation tools. It’s even worse if you’re a freelancer. What to do? Ignore any new business that comes along demanding different file format requirements? Is there any way to reduce or remove the translation burden for the small company or the individual?

When costs and expertise make an internal process infeasible, it certainly makes sense to outsource. But when needs are variable or one-off, simple outsourcing is not enough, you need an on-demand model. So could CAD translation be turned into an on-demand service? One startup in particular, Online CAD Converter, is aiming to do just that. They are testing the service now, with an introductory offer of 3 free translations to give the concept a whirl. For some of the same reasons that the drive-thru (European readers will have to use their imagination) at your local burger palace provides a low friction meal on the run, no kitchen or pre-planning required, engineers could certainly use a similarly accessible and hassle-free translation experience. Given the right format support, it might be the next best thing to tater tots. So what do you think?

  • Wow you put my observations into words. I started my own
    engineering business two years ago with the expectation of purchasing a 3D CAD
    system within the first 6 months. I was shocked to find out how many CAD
    systems there were. Structured modeling, direct modeling, some that offer both
    with different pros and cons. I have many clients and some may be a onetime
    deal. Luckily my work has not required more than AutoCAD so I have been able to
    postpone a 3D purchase. I was surprised to learn that even with the translators
    being much better than 20 years ago moving from one system to another turns
    structured models into dumb models. Some may argue that structured models are
    not worth the effort but if a client were to provide me with one to be updated
    and I did not have the same CAD software and rev level I would return a revised
    model that is no longer structured and the history gone. Yes as you go down the
    supply chain you will likely have many clients with a small portion of your
    revenue coming from each. So how can any small firm maintain 3 or 4 CAD systems?
    The translation service would be a good thing and perhaps force the main CAD
    developers to get their systems to speak to each other.

    • Thanks for recounting your own experiences on translation difficulty, do you prefer to use parametric based design or direct editing and why? I’d be interested to know if the translation service mentioned in the article proves to be relevant to you.

    • Joe, I agree that a small firm cannot maintain 3 or 4 CAD systems; larger ones sometimes have to, for example we have one customer who has to save their work to NX 9.5 with NX drawings for each part. That’s something we can’t provide, they HAVE to buy a seat of NX to comply, if they want the business.

      I am not as hopeful as you are about translation services forcing the big vendors to get their systems to speak to each other. It’s based on capitalism and competition; they continue to implement new nuances like CATIA’s Publication Data, which now must be understood and translated if it is being used by your customer.

      It’s a constant battle. As new features emerge, translation becomes even more complex, yet tools for translation and feature recognition continue to improve. Like road construction, it will always be with us.

  • Good article; it is indeed a daunting landscape for small firms to navigate these days. It is good to remember that nearly all translation tools will end up giving you a “dumb” solid (unless you have ~100k to spend), but all the major CAD systems also come with feature-recognition tools, so in many (not all) cases, you can get your features back. This, combined with the increasing access and use of direct modeling tools (no feature history) makes working with a “dumb” solid less of an issue. Another positive is that the barrier to entry for translation and repair has dropped significantly in the past several years. Small job shops can have the same tools that many of the bigger firms use. TransMagic (the company I work for) offers a suite of affordable tools and currently also offers a two week eval, so you can try before you buy. The proprietary formats and big CAD politics will always be with us, but the way I see it, things have never been better for the small job shops, who can now at least afford to work on a level playing field with the larger firms.

    • Thanks for mentioning Transmagic, do you think a market exists for on-demand translation TAAS (Translation as a Service) or do you think the small shops and freelancers can afford the current level of traditional license fees?

      • Well, first of all, the link above to the Online CAD Converter is notable in that it does not translate out to true CAD formats (at least not yet) – all of the output formats they list are polygonal formats, ie, non-brep.

        Sure, I think a market exists for online translation; as the article states, this is a huge issue. To me, the question there is, what tools are available, how good do they work, what do they cost, and how well are they getting the word out?

        And, yes, at the same time the small shops can afford the current level of traditional license fees. TransMagic Pro for example currently costs $1450 per year; for that you get an automatic repair tool, can read any major CAD file, output to brep (true CAD) neutral formats, and have access to visualization and markup tools. That may sound like a lot for a company that is working on thin margins but one also has to weigh the time lost due to bad translations, manual repairs, and remodeling parts from scratch.

        Throw into that the cost of having to turn down business because the customer has files in a format you can’t open. It’s also sometimes true that not everyone using a CAD seat these days actually needs a full CAD seat, they might get by with a less expensive and easier-to-use visualization and markup tool.

  • Ross Bernheim

    Ed,
    Proprietary document formats that don’t translate are not confined to CAD. Other types of graphics programs, video programs, accounting, and word processing programs also do this.

    CAD has many programs, but no totally dominant program. To see what happens when there is a totally dominant program, look no further than the word processor segment. (I know that word processors are usually bought as part of an office suite.)

    Microsoft Word is the dominant program. You can argue about if they got there via underhanded or illegal means, but that is water under the bridge. They are there now.

    Microsoft creates a new document format for each new edition of Word. They also set it up to automatically save in that format. Of course the older versions do not read or use the new version so as soon as you have one person using the new version of the program there is pressure for everyone to upgrade.

    Other programs such as Libre Office and Open Office among others can read Microsoft Word documents, just not the latest version until they can come up with a translator. Even then there are a lot of undocumented things in the new file format.

    I had the lovely experience of having someone do my resume a few months ago. She is a heavy Microsoft Word user. I lack that program and use Libre Office. Much ‘hilarity’ ensued.

    She was doing things like adjusting word and letter spacing and other very fancy and subtle formatting in Word. First step was showing her that such ‘desktop publishing’ was not a good thing to do. As well as it being wasted effort as the document scanners that companies use won’t be impressed by it and neither will hiring managers or recruiters who spend mere seconds scanning your resume. Next was alerting her to the security issues with the Microsoft Word document format. Then there was the issue of converting to PDF format.

    Net result was she learned a lot and now uses Word for its basic word processing functions and forgoes the fancy non-translatable functions. But now she is able to do resumes that look as nice with less effort and avoid the security issues and her documents translate to other programs almost perfectly every time.

    You might ask, what if we had open standards. Most of these problems could be avoided. Again Microsoft provides a great example. Libra Office, Open Office and several other programs share a common open format that was becoming an ISO standard. When several foreign governments decided that they wanted to go to an interoperable ISO format, Microsoft had a problem. Their solution was not to use the already completed format, but to create their own and ram it through. Now they can claim to use an ISO standard format. The problem is that it isn’t fully specified so others can not implement it. Even worse, Microsoft Word does not fully comply to their own standard.

    Will sanity and rational thought prevail and give us document formats that can operate across our may programs in any field? I’m hopeful, but not holding my breath.

    • An interesting example with MS Word, Ross. You’re right that no one vendor has managed to secure dominance in the market to the level that Microsoft has with office applications. Perhaps what keeps the CAD market diverse is the amount of specialization in many of the products? Or perhaps the longer purchase/upgrade cycles?

      When MS moved the document format to Open Office XML with 2007, that was supposed to end the days of proprietary binary files for Office and the translation headaches you mention. While the content and format is plainly readable in the OOXML, how Microsoft renders that XML in Word is still proprietary. That’s why Google Docs, for example, still do an awful job with Word docs that use more than the most basic formatting – despite the open file storage format.

      It’s that problem for which some in the scientific and academic communities argue for the use of document markup languages like Markdown or LaTeX, to ensure consistent communication of content and formatting. The problem is working on a document in a markup language, is that is extremely user unfriendly – more akin to programming than the visually rich WYSIWYG model of modern Word processors.

      STEP 242 in many ways is supposed to be the OOXML of CAD, provided applications were converted to read and write in it directly. But the standard is so enormously complicated, by the time it sees the light of day it may very well be obsolete. The short of it: OOXML is a hard enough problem for formatted documents – doing the same for 3D geometry is at least an order of magnitude more difficult. So it seems we may be stuck.

      • Ross Bernheim

        So instead of a proprietary ‘open’ format used to disadvantage the competition the alternative is an open format by committee.

        Step 242 is doomed by the inertial of the committee as well as trying to include every single feature that the many proprietary programs can throw at it as they try to gain advantage in the marketplace by adding new proprietary features and match all the other proprietary features of the other competitors.

        The result is ever larger more bloated CAD programs with user interfaces that are ever more difficult to navigate and use.