I’m uploading a small part of my model where intersection fails at tolerance = 0.005
I have similar intersections to handle in this model and so far 0.005 works best except at this spot. If I change the tolerance to higher or lower the problem disappears at this spot, but will appear at other spots.
ps I uploaded this in Windows category as it is not really a gh releated issue. Doing this in plain Rhino gives the same issue.
intersection-fail.gh (42.9 KB)
your Breps are quite far from origin - if there is no reason for this - use geometry closer to world origin.
most common tolerances have a 1 as last digit…
maybe a 5 at the end is a bad choice as rounding takes place between 0.00499999 and 0.0050000 … ? just a guess
which rhino version do you use ?
hope that helps - best -tom
I don’t think they are far enough for causing rounding issues. When I move them close to the origin, it still fails.
like I mentioned, when I use other tolerances (0.01 or 0.001) this one doesn’t fail but other similar trims are failing. And since there over 500 of these trims to make it’s become a weak link with this issue.
I suspect it has to do with the fact that the kink in the object I try to trim with is exactly at the horizontal surface that connects into two fillets. In this case modeling precisely seems to work against me.
this was made in 6, but it fails the same in 7
ok i managed to reproduce your error.
there was a bug with intersect in Rh6.
why not work with 0.001 tolerance ?
if you want to stay with 0.005 another workarround is to deconstruct your brep first and join it after trimming:
if your still not happy you might include an (AT)pascal AT=@ in your next answer. - i can t help any further - kind regards -tom
could be but it fails the same in V7
as mentioned, then other trims are failing. there isn’t a single setting that works for all trims
thanks I’ll try that
Usually it is best to stay with the same tolerance for an entire model, or if it needs to be changed make it larger only.
thanks @Tom_ it seems to work well with your suggestion although this part of the definition has become an even greater bottleneck. Even before that it took 4 minutes, but now it takes 8 minutes. But that’s better than failing 
wouldn’t that need to be ‘smaller’ only? In other words if my abs tol is 0.01 then there is no problem lowering that number temporarily. Or do you mean make it larger and then stick with that larger tolerance?
No, new tolerance should be larger than the previous tolerance only.
Geomety created with tolerance set to 0.001 may have gaps as large as 0.0099. If the tolerance is changed to 0.0001 then the existing geometry may have gaps larger than the new tolerance which is can cause problems.
If a tolerance is changed to a larger number, for example from 0.001 to 0.01, then the gaps in geometry created with a tolerance of 0.001 will be smaller than the new tolerance.
I think we mean the same, so either change it to a larger value and stick with that or temporarily make it smaller and change back to the original setting.
