CGS versus Brep

Dear folks,

out of curiosity as a (systems) programmer, i'm currently evaluating a number of (solid) modellers and their implementation techniques. What i am seeing is the almost absolute dominance of boundary representation schemes. I would expect to see more CSG based modellers since they are far better suited for CAD/CAM IMHO. Shelling, filling and rounding etc. can also be expressed in CSG so why the dominance of B-rep? And are there real CSG modellers around still even? BRL-cad comes to mind but it comes over to me as ancient and with a horrible user interface.

SolidWorks, CATIA, IronWorks, Unigraphics NX to name a few all seem to work on the ParaSolid kernel with different user interfaces and tools and thus are basicly boundary based though annotated on how the solid was created.

Well, since you folks are working with CAD on a daily basis, do you think that a non B-rep CSG based system could be feasable or would even be desirable if confined to industrial design and manufacturing i.e. not for pretty pictures but for machinable parts? Would NURBS free form primitive support be crucial? or is it overrated in this field of work?

With regards, Reinoud

Reply to
Reinoud Zandijk
Loading thread data ...

Reinoud,

CATIA is not Parasolid based, they've got their own, but anyway... CSG has serious limitations. Even without going too far, a simple fillet is a big task when you want to construct it from primitives. B- rep handles complexity well. Also, there are tasks when any system can only approximate, like intersection of free form surfaces, etc. Today's practical CAD applications with robust solid modeling are mostly feature based, because the natural thing is to build things from / with features, not from primitives. If you want free form stuff, then your primitives are not so primitive anymore. For industrial strength applications one needs something robust, not something what looks good in theory. Parasolid is well developed and it does more than just a kernel, such as preparing geometry for graphics, etc. You can sign up to get a developer's kit, develop your app around it and pay a modest license fee on it (Same with ACIS). Before I forget, it is also well proven. But if you want something different look at OpenCascade.

Regards Attila

Reply to
bertok

You are right... most real CAD systems are based on the BREP with different user interfaces. Excepting interface changes and certain modeling algorithms (like SolidWorks Boundary and Surface Fill), they are mostly interchangeable. And even these goodies are modifications beholden to the BREP. And when it comes to NURBS, there are still unrealized applications.

The only thing I could find on a quick search about CSG based modellers is that ' constructive solid geometry is performed on polygonal meshes'.

Please let me know if I missed something, but as described meshes are crap dead-ends for manufacturing. Give that up - they are old, dead technologies, way behind the curve.

The thing that bugs me as a professional product developer is that the BREP, as a product, feels like a dead end. It's good - we get our jobs done, don't get me wrong about that. There is a lot of history behind that statement that would take a long time to get into. But after being at SolidWorks World 2008 (and prior history) for a wild night where I was meeting with a number of folks from a number of established and startup companies it was clear that most CAD we see today is being driven by folks pursuing the still unrealized aspects of the Computervision dream. At the core of everyone's development is the BREP, and they are all chasing the same dog. As someone who is very conscious of the BREP, I know that most of our modeling pain come from the limitations of the BREP.

I can say this... 20 years from now we won't be talking BREP anymore (excepting legacy applications). There are new techs I've been hinted to, and there better be lots of new techs that "I am too small a player" to be hinted to. The BREP is near its end. My gut says that in five years we will see the first real commercialization of the new paradigm, and in 20 years we will be saying BREP... what's that???

Reply to
Edward T Eaton

Thanks a lot for your comments Edward,

CSG modelling is not a new paradigm, it was implemented in PADL-2 before CATIAv1 was even out but its not used that much since rendering CSG models is inherently slow; its more like raytracing. A CSG description is purely mathematical; every surface patch is mathematically described and when combined its mathematically soundly connected. Some applications like SolidWorks (but also Blender) seem to do boolean operations on polygon meshes whereas IMHO BRep/polygon meshes should only be used for displaying.

I think you might be right :-S

Also in SolidWorks? What problems do you run into?

Would you care to elaborate a bit on that new paradigm? if by email if you prefer?

With regards and thanks in advance, Reinoud

Reply to
Reinoud Zandijk

Well, without a form of macro's it sure would be a big task, but i was assuming that a fillet on a CSG based system would be implemented using CSG modelling but with the UI of today's fillet operations i.e. automated generated operations that together perform the fillet.

Thanks for the link to OpenCascade! I'll try it out too though i think its BRep too. True, Parasolid looks like well developed and it does do a good job for todays systems.

With regards, Reinoud

Reply to
Reinoud Zandijk

???????, 21 ????? ??? 2008 ?., 8:36:03 UTC+3 ????? ??????? Edward T Eaton ??? ?????:

almost 10 years later this is kinda not true

Reply to
babochenko.denis

I am super dissapointed to say this has not aged well! I am doing a course at Loughborough Uni and found this comment looking up the difference betwee n Brep and CSG because apparently its still relevent (my god the lecture no tes are old though). Either way, this discussion has been useful to me, so thank you for having it 13 years ago!

Reply to
Daniel Timmerman

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.