That's not going to be a short one, but it's certainly possible. The trick is that the segments at
different angles all repeat at different cycle lengths. E.g. the horizontal part at the middle of
each "curve" repeats one curve over, so its on-off cycle total will be simply the distance between
sets of "curves." But the adjacent segment at a slight angle repeats in some "curve" farther off
and up or down some; and that relationship varies with each segment.
I would do this by drawing a field of portions of regular polygons in the relationship you want,
exploding one set in the middle of things, and extending the Line segments to find out where they at
least come close to meeting the same point on another pattern somewhere. "Close enough" is going to
be a factor in determing how you tweak their angles to get something that repeats properly. Then I
would break the Lines with no gap, at the intersections with adjacent segments. If you draw all of
that with a logical pattern-generation origin point at 0,0 [such as the nominal center-point of the
"curves"], you can read the start point coordinates, angles and on-off lengths directly from the
resulting Lines.
The attached might get you started. It makes 1-unit-diameter "circles" in the form of 24-gons
[which are not quite regular, because of the nature of hatch pattern definitions, but are very
close], in a triangular grid with 1 unit spacing between. A portion of that [not all of its
segments] could get you going on your outermost "curve." [I don't remember off-hand whether some of
those lines of code define a portion on the bottom of some circles and the top of others -- if so,
there would be further adjustments to make than just limiting the number of lines of code used.]
And since your pattern of patterns looks square/diamond-shaped rather than triangular, the angles
[and therefore the on-off cycles] will need to be different to get the segments to fall in the right
places on the repeats. But you can at least see how it goes together, if you analyze it in relation
to Help's descriptions.
You would then need to repeat the definition of the one "curve" six more times, changing the
starting locations and on-off cycling for each. But if you keep them all based on a 24-gon
approximation, the total of the on and off for each angle involved would at least be the same for
each "curve." You could also simplify the smaller ones into fewer segments, but that might depend
on the scale you expect to use it at, and how much visible segmentation you can stand.
You might also experiment with a different number of segments in the curves. Your diagram looks
like 24-gons would be right, but it involves half-length segments at the ends of at least the longer
curves. Given a 90-degree sweep for each curve, 20-gons, or 28-gons, or 24-gons with a *vertex* at
the middle rather than a segment mid-point, would end with complete segments. The first case would
only require five lines of code to define a curve, and the last one six lines, rather than the seven
lines needed for what's in your diagram or for the 28-gon variety.
--
Kent Cooper
kdepfyffer wrote...
I'm trying to create a hatch pattern and don't really know how to handle this one. I can create
simple ones but since these lines don't really line up I don't know how to handle it.
attached is a pic - if you can give me some pointers or help me out please let me know.