|
|
Le_Forgeron <jgr### [at] freefr> wrote:
> Le 12/02/2016 17:50, Bald Eagle a écrit :
>
> > This gives rise to a further refinement of slanting the puzzle edge and
> the
> > locking tab to smoothly follow that angle, which shouldn't be too bad o
> nce I get
> > back into the swing of things.
> >
> > Then I just need to come up with some kind of formula for preventing tw
> o slots
> > from colliding with each other if they get generated too large or too c
> lose, and
> > wind up constricting and cutting off that corner.
>
> I hope you want something like the attached picture (well, if I got it
> right).
>
> The pattern is using 2 exclusive zones, with security margin to avoid
> problem.
>
> Everything with a cubic_spline (and sphere_sweep).
>
> cubic spline is required for: smoothness and going via the point.
>
> In the triangular area, there is at least 4 points (but a fifth can be
> added). the first point must be somewhere in the green part.
> the second point is along an offset diagonal, and so is the third point.
> an intermediate point between second and third can be added, as long as
> it is between these points on the green-yellow axis, and nearer the
> centre of the tile (without entering the no-go zone of blue diagonal.
> the last point is back on the axis, somewhere in the yellow line.
> then we can reach the corner point.
>
> For a non-offset cross, this is all that is needed.
>
> For an offset cross, as done by the vertical line, we need more points:
> * on the horizontal line, we need to be sure that the line is flat for
> the segment inside the small diamond at the centre, so we need an
> intermediate point after the yellow zone, and before the corner point,
> possibly near the intersection with the blue line and the horizontal
> when entering the diamond.
> * on the vertical line, we need to negotiate a corner with the spline,
> so even more points are needed, including one very near another just
> before reaching the horizontal segment (cubic spline has a small memory,
> so we need to make it forget a bit of it when joining the horizontal
> segment)
>
> That's for one crossing, it can obviously be used alternatively for the
> vertical and horizontal offset on a regular grid.
>
> Also attached the scene for it.
another option I hadn't thought of earlier would be to populate the interior of
the area you're cutting into pieces with rectangular regions of varying
dimensions (within a certain range) and then "Connecting the dots" with some
amount of turbulence
A rule could be applied that any interior edge over a certain length has to have
a tongue or notch.
If you limit the minimum perimeter of a given region, edges below that length
can be assumed to be an offset cross.
Regards,
A.D.B
Post a reply to this message
|
|