On Fri, Jun 29, 2012 at 12:46 PM, phoenix <284281...@qq.com> wrote: > Hi, > In the last couple of days I have implemented the b-rep conversion function > for superell. When n <= 2 and e <= 2, it produces the exact same result as > we want, but when one of them is greater then two, they are slightly > different.
Different from the implicit form or different from examples elsewhere? > (This may result from the non-smooth surface of the super > ellipsoid when n or e > 2, as the NURBS surfaces always tend to be smooth.) Can you post a few pictures illustrating the problem? If you need sharp corners, I think there may be a way to do that (if nothing else using multiple surfaces, but there may be other options as well.) > And I also looked into other primitives, and have some questions. As for the > 'grip' primitive, is it something like half? I found it does not show in > raytracing. > So, in what situation do we usually use this primitive? And for > what purpose? IIRC, grip is used to describe relationships between other primitives - it doesn't have a shape itself. So it's not a candidate for NURBS conversion. > Besides, I'm also quite confused with the 'cline' primitive. > Its detailed name is FASTGEN4 CLINE, and FASTGEN4 is another data format. > After I raytrace this primitive in different views, I still cannot imagine > what its shape should be. So, could someone please figure it out? The best I can offer is the following from the documentation of the system from which cline originally comes: http://bzflag.bz/~starseeker/cline.pdf If the raytrace results of cline are view dependant (I don't recall for sure) then it cannot be represented properly as a NURBS brep in any case. > At this time, most primitives already have implemented b-rep conversion > functions and the functions works quite well, and I will still test the > results and modify the code if there's some problems. And, what else can I > do besides the testing and modifying? Could someone please give me some > suggestions? Well, if you want to undertake the next phase (going from unevaluated CSG trees to evaluated ones) the problem that needs to be addressed is the so-called "surface-surface" intersection problem. Starting with a simple example - let's say you have two spheres, one of which is being used to subtract a "slice" out the other one. Right now, we can represent the two individual spheres as Breps, but we can't express the shape that results from the subtraction. To do so, you need to describe the intersection of the two NURBS surfaces representing the spheres (a NURBS curve, in this case, although it might be a point or a surface in some cases) and create new NURBS surfaces corresponding to the first sphere without the "carved out" part, and the "negative surface" from the second sphere that goes into the hole in the first. The combination of those two surfaces describes the "evaluated" boolean of the two spheres. There is a fair bit of literate on the topic, and a variety of possible approaches, so if it sounds interesting you can look it over and see what you think. Some quick links: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.42.181 http://www.cs.berkeley.edu/~hling/research/paper/curve.htm http://www.cs.berkeley.edu/~hling/research/paper/intersection.htm http://dl.acm.org/citation.cfm?id=1364937 (intersection discussion it towards the end - we don't currently use GPUs but often the algorithms are useful even without them.) http://www.idav.ucdavis.edu/publications/print_pub?pub_id=80 Cheers, CY ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ BRL-CAD Developer mailing list brlcad-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/brlcad-devel
