mmyoung Posted August 5, 2009 Share Posted August 5, 2009 (edited) We've discovered that the Unfold Surfaces Tool fails at a crucial point in a sculpture we are fabricating for an artist here. I've included two .vwx files. "3 turn spiral A.vwx" is the one we fabricated from. The other file, "spiral mmy 01b.vwx" is another attempt with the same geometric result (unbuildable). The red unfolded surface in "3 turn spiral A.vwx" is the one that pulls away from the structure about half-way up, though the other two surfaces seem to be fine. Attached you'll see two photographs ("unfold error 1.JPG" and "unfold error 2.JPG") from a complete set of documents showing the entire process, in which we attempt to fabricate a small-scale mock-up for 30' tall sculpture. Our complete documentation set shows exactly where the part failed, and we can send it if it will help. The last photo ("spiral fabbed.jpg") shows a completed spiral in which the piece that failed was re-fabricated by hand (tedious and difficult even for our crew). The final, full-scale piece cannot tolerate even minute irregularities, however, and building it at full scale would be too risky, so we have not yet taken the project all the way to completion. Any assistance or insight would be very welcome! Has anyone else had any experience with the Unfold Surfaces tool like this? Edited August 5, 2009 by mmyoung Quote Link to comment
islandmon Posted August 5, 2009 Share Posted August 5, 2009 I don't have VW09sp4 so can't access your files. But the project looks like a wonderful challenge ! Since your errors appear after the first pi rotation , my gut tells me that you may have a common resolution issue involving the compounding of trig. rounding errors. Possible solutions : 1) Program to maximum decimal accuracy. 2) Use radians. 3) Set-up 'key frames' via Working Planes at the specified 180? (3.141592654r ) rotations to check the rotational accuracy. 4) It may be possible to segment the curves into shorter lengths.. then add the curved surfaces in the unfolding. Thnx for sharing ... Quote Link to comment
mmyoung Posted August 7, 2009 Author Share Posted August 7, 2009 Islandmon, thank you. I was hoping you would look at this. Great suggestions, of course! We will examine it again in light of these, and I'll re-post when we find out how this works it out. Do you have time to take a look at it if I backsave it to ... (I forgot what version you like)? Quote Link to comment
mmyoung Posted August 7, 2009 Author Share Posted August 7, 2009 Would you happen to know how to tweak the floating point accuracy of VW?? Quote Link to comment
Ray Libby Posted August 7, 2009 Share Posted August 7, 2009 File menu>Document Settings>Units...>Rounding Precision>Decimal Precision: Quote Link to comment
propstuff Posted August 7, 2009 Share Posted August 7, 2009 I was under the impression that the underlying calculations were always done to the max decimal accuracy and that rounding only affected the display of numbers. (?) Quote Link to comment
Ray Libby Posted August 7, 2009 Share Posted August 7, 2009 True, but when entering decimals I would think you would get increased accuracy. Quote Link to comment
islandmon Posted August 9, 2009 Share Posted August 9, 2009 When it comes to complex 3D work involving rotations, vectors, & arcs , I always program to the full 10 decimal places . This is really the only way to get the elements to close sufficiently ( as seen at max zoom ). Quote Link to comment
propstuff Posted August 9, 2009 Share Posted August 9, 2009 When it comes to complex 3D work involving rotations, vectors, & arcs , I always program to the full 10 decimal places . Do you mean that (for example) if you want something to be 100mm long you actually enter the value as 100.0000000 ? N. Quote Link to comment
islandmon Posted August 10, 2009 Share Posted August 10, 2009 Yes .. that is what I do .... although 100mm seems a fully rounded special case ... that number could just as easily be 100.0012345678 ... but you would never know it unless the units were set to read more than two decimal places. Now that may not be a big deal for those who do not require closure... but I like my theoretical models to be as close to perfect as possible when I zoom to the limit. Try setting Units to 0.00 then mess around with a few object placements using the both manual placement and the OIP to fix the points. Now reset the Units to 0.0000000000 and see what you have wrought ... you may be surprised. Now duplicate each object and rotate around the XY and Z axis ... small errors begin to compound into significant displacements. All i'm saying is that with compound rotational displacements in 3D space it is essential to check the math prior to each iteration. That is ... if you want to avoid compounding the small XYZ misalignments between the parts. Quote Link to comment
mmyoung Posted September 10, 2009 Author Share Posted September 10, 2009 Islandmon, It did turn out to be a rounding error! Thank you. At least, that's the only explanation that survived our investigation, and it bears all the stigmata. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.