bbox experiment re suqare finger joints

17 views
Skip to first unread message

Loïc Fejoz

unread,
Dec 31, 2015, 5:20:14 AM12/31/15
to IceSL
Hi all,

I have try to use the bbox function to automatically do square finger joints on any two parts.

https://github.com/loic-fejoz/loic-fejoz-fabmoments/tree/master/autojoiner

Globally the results is really really slow!
I know my video cards is not that good but it is barely usable in that experiment.

I first made a script that was using more unions. It is the one used for the screenshot: the result is correct but (even more) slow.
Then I used differences: it is faster but the result is bulky on tangential faces.

Note: I have downloaded the latest version on MS/Windows and it seems it is of lesser quality (editor, missing operations on vector, print and console, editor scrolling, etc). I will try to later write a list.

I will be pleased to know what you think of it.

Regards,
Loïc

sylefeb

unread,
Dec 31, 2015, 10:05:20 AM12/31/15
to IceSL
Hi Loic,

Great tests! I remember fixing a bug for the next release that will make it faster in a number of cases.

However, the performance will generally be much faster if you call bbox on a mesh (or tesselated such as sphere, cube, cylinder) than when you call it on a CSG compound. That is because to compute the bbox on a CSG shape, IceSL has to resolve the CSG at a reasonable resolution. If I expose the resolution (think 'precision') then you could play with the speed/precision tradeoff. Maybe that would help?

Best,
Sylvain
Reply all
Reply to author
Forward
0 new messages