The odd thing is that decreasing the tolerance does not uniformly improve the output. And going down to 0.0000001 produces the same as 0.000001. Effectively 0.00001 + 0.000001 is the best, but still not ideal.
I curious what I should be doing to get the best output here.
Hi @bigjimslade . What Units are you working in that you require 0.0000001 tolerance?
Metres?
If Units set to mm then try setting tolerance temporarily to 0.1 and you will see that Make2d runs much faster and creates fewer line segments which makes the 2D version more editable and usable. Michael VS
I am working in inches. There are details that are quite small. I have to use 0.00001 to build the model. So that was what I did, unchanged, initially. However, a lot of lines that are needed fade out. The object here is 5-1/2" long but some of the lines are very close, down to 0.001.
At your hint, I tried going down to 0.01 and work may way up to 0.00001.
0.01 looks horrible.
0.001 actually looks better.
0.0001 starts to look worse. (However, there are details at 0.0001 that do not show up at 0.001)
Maybe the answer here is the tolerance requires turning for best results—unless there are other suggestions. Maybe a combined image using 2 settings is necessary.
My theory on tolerance for make 2d: You cant complain about messyness in the result that is off by a couple of multiples of tolerance. So if there are gaps between curves of size tolerance , or a curve that is partially missing but there is a nearby curve within tolerance, then these are acceptable results at that tolerance. The idea behind this is that the result is only going to be viewed certain limiting scale, say its going to be printed on a certain size of paper. At this intended scale errors of certain size are just not visible.
Now all these grand goals aside. I am sure there are bugs and I don’t think there has been much testing of how the results very as the tolerance is changed. So I am interested in what you find. I’d like to see models.
Finally a general comment about tolerance settings. If you are using 1e-5 as a tolerance setting your object and everything it interacts with should probably have absolute coordinates less than 10 or maybe 100. Otherwise, dicey calculations ( technically known as poorly conditions) will be susceptible to round off error in numerical computations.