CGAL users discussion list

["J.L.M." <>]

John,

thanks for your help. I'm reasonably confident that my models do not 
violate the restrictions. I had run into those early in my work, and the 
assertion clearly points out when you are violating it. Attached is a 
picture of my model.

How are you using OBJ models in CGAL? I see that CGAL supports writing 
.obj files, but not reading them back in (unless I missed the 
documentation). The only format I see that CGAL supports for reading is OFF.

The erosion should produce exactly two lines, attached at one endpoint 
at a 90 degree angle. I am testing the tight passages scenario. Indeed, 
when I use a smaller cube, I do not get an assertion.

I do understand that there can be translation if the shapes are not 
properly positioned about the origin. My models always include the 
origin on their interior.

On 12/28/2010 09:31 PM, johnzjq wrote:
> Hi, J.,
>
> I have tested your models. At first, I came across the same 
assertion! Then, I examined your model and did some modification, 
Finally, succeed.
>
> 1 The models you provide probably trigger some of the restrictions of 
> MinkowskiSum.
>  (S29.3 
http://www.cgal.org/Manual/latest/doc_html/cgal_manual/Minkowski_sum_3/Chapter_main.html)
> I converted the model to the triangle based OBJ model using 3dsmax 
and there is no assertion appeared any more.
>
> 2 The l_shape is too small to produce any results after "erosion". I 
> scaled the original model by 4 times (or 5) and get the result as 
attached.
>
> (Carelessly, I translated the coordinates of the original model. 
Please try with the new models)
> Hi,
>
> Many thanks for your detailed explaination. And your suggestions are really
> helpful.
>
> I have implmented the code I attached in the post "Any idea about Minkowski
> difference?" and it works fine on many complex CAD models (which should be
> 2-manifold) , but a bit slow.
>
> Regards,
>
> John

Attachment: l_shape.jpg
Description: JPEG image