Hello Ole,

Thanks for the message!  Ah yes, the intricacies of the pernicious g- 
stl expoter..  If you can provide as small a sample of geometry as  
possible that reliably produces a g-stl crash, that would be  
incredibly helpful.  Problems with the converter are generally  
fundamental to the tessellation method it uses, but there are various  
instances of bad geometry and an occasional bug that can really only  
be fixed when they can be easily isolated and reproduced.

> I find BRL-CAD to be just the tool I need for my 3D modelling. But I
> need to export my models to STL and for that I am using the g-stl
> utility.
>
> I am on Windows XP with version 7.12.6.
>
> I have a question with regards to the g-stl and its functional  
> stability.
>
> I have had some problems with it myself:
>
>   1. I takes way too long with small relative and absolute tolerances
> (parameters -r, -a, and -n).
>   2. It can suddenly crash in the middle of exporting. This usually
> happens if tolerences are low (0.001, 0.0001).
>
> What is the best way to use g-stl? I need som good SLS/SLA prints  
> of my model.

If the problem is bad geometry, you may have better luck avoiding g- 
stl altogether (which attempts to validate/repair geometry) and  
instead use our bot_dump tool (which does no repair).

> - Should one use combinations only, or should one use regions only?
> Does it matter?

There are many configurations that will slow export down considerably.

As for combinations vs. regions, you shouldn't exclusively be using  
"only" either one of them.  They have rather specific meaning and the  
construction will affect how an object exports.  A region is sort of  
like a part, in common CAD vernacular, and a combination/group is  
akin to an assembly.  Basically, a region is where you identify that  
a particular shape has become solid.  You use combinations to group  
those solid objects together.  The main confusion is that you can  
also have combinations *within* a region (and regions themselves are  
technically combinations), but below the region level they are merely  
shapes or templates and don't occupy physical space yet.

> - Any other good ideas for the boolean-structure of the models in the
> .g file? Is there some configurations that always will slow it down?

I'd recommend reading Principles of Effective Modeling at http:// 
brlcad.org/wiki/Documentation if you haven't yet as that covers lots  
of best practice issues with hierarchy construction of models.  The  
OED command tutorial also covers lots of detail in the process of  
explaining how object editing works.  There are lots of ways that you  
can make it be horribly slow.

> - Can such combinations/regions contain other regions (would it
> represent a possible exporting issue if they do)?

Regions within regions is considered a modeling error.  You can,  
however, use/have simple combinations that do not contain regions  
within a region.  You can group as many regions as you like within a  
combination, though, and make combinations of combinations as needed.

> - Can somebody explain or recommend on the use of the arguments -D,
> -r, -a, and -n?

That's rather specific to your geometry.  Some types of geometry are  
very sensitive to -n normal curvature, others are more sensitive to - 
a absolute tolerances because of a mix of large and small features.   
It's generally best to start by just changing one of the three and  
take it up/down orders of magnitude and see how it affects the output  
until you get a good feel for how a model is responding to different  
tolerances.


> The extra arguments for setting tolerances seem necessary, since too
> few triangles will be created otherwise.
>
> - What is the functional stability of g-stl? Is it working, is it
> buggy, should and can I rely on it for exporting my MGED database
> files?

If you find a bug, please do report it.  The tool is in production  
use (on a nearly daily basis) and is generally considered pretty  
stable.  The algorithm it uses for tessellation, however, is  
unreliable and can cause really long run-times and a variety of  
unreliable results if your input geometry isn't clean.

> - Is there anything to be gained by using BRL-CAD+g-stl on linux
> instead on windows?

Not particularly other than you can more easily script g-stl on Linux.

> Regards,
> Ole Buus

Cheers!
Sean

p.s. A Windows binary for 7.14.8 was just made available on Sourceforge.



------------------------------------------------------------------------------
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
_______________________________________________
BRL-CAD Users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/brlcad-users

Reply via email to