On Tue, 4 Jan 2011 10:11:48 -0500, Mark wrote: > (Last discussed here March 09) > I have done some more work with creating solids from gcode. > > Creating a line to show the path of the tool isn't bad. I'm not sure > that > arcs are always created with the correct rotation, but other than > that I > think it is accurate. > > Creating solids can be a pain, as opencascades boolean ops are > finicky. > > Creating *accurate* solids is a real challenge - if the tool is > moving in z > as well as x or y, the shape removed from the stock can be quite > complex. I > can use occ's Hidden Line Removal to get the silhouette of the tool, > but I > get extra stuff that I need to throw away - such as overlapping > lines. I'm > having a lot of trouble solving this. > > Another issue is that boolean ops are slow. If you have one for every > line > of gcode, and each one takes .5 seconds or more...! > > Source at > http://code.google.com/p/cam-occ/source/browse#svn%2Ftrunk/src/g2m > > Screenshot > > http://cam-occ.googlecode.com/svn-history/r136/wiki/images/face-sc-sq-step.png(the > steps shouldn't be there) > > There are a few other people working on the problem of modeling > material > removal. Unfortunately it seems that our approaches are different > enough > that merging the projects would be difficult. > > If anyone has suggestions for solving these issues, let me know! > Mark > On Mar 4, 2009 3:20 PM, "EBo" <[email protected]> wrote:
I do not remember what I wrote a year and a half ago, but here are some things which might help. The are a number of boolean operation, or related ones like mathematical morphology, toolboxes wich should be useful. The one I remember off the top of my head is GTS <http://gts.sourceforge.net/> which might be helpful. If I had time I would also look at the Visualization Tool Kit (VTK) <http://www.vtk.org/>. VTK is extremely well supported and has an open source license (as well as commercial ones). As for your step problem, it looks like you are linearizing the curve, or subsampling a linearized path. There are two ways to deal with this. Make the steps finer (which just makes the problem smaller), or use swept surfaces and cauculate the surfaces directly. If you want to understand the low level stuff on how to do this efficiently I would recommend Gerald Farin's book on Curves and Surfaces. I just had another thought, if you take the local minima, then you can connect them and approximate the surface from what you already have. That will take some low level tweaking, but probably the fastest. Hope that helps. EBo -- ------------------------------------------------------------------------------ Learn how Oracle Real Application Clusters (RAC) One Node allows customers to consolidate database storage, standardize their database environment, and, should the need arise, upgrade to a full multi-node Oracle RAC database without downtime or disruption http://p.sf.net/sfu/oracle-sfdevnl _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
