No torsion in frames along a curve

Replies to This Discussion

Permalink Reply by martyn hogg on May 25, 2015 at 12:45pm

Have you tried to align the planes with the guide vector being Z?

Perhaps it's something to do with the way you are constructing or moving the lines to the planes?

If you post your def we can have a look

Permalink Reply by Stijn Rosseel on May 25, 2015 at 1:51pm

I've tried to align the planes with the z-vector but no succes there.

Only thing that seems to be working relatively well is the solution below (purple box)

In the right panel I'm keeping an eye on the difference in z-coordinates (0.0 is what I want).

The only problem is that sometimes the plane won't orient right and it'll flip, I have no clue as to why this happens.

Is there a way to control or fix this issue?

I also included my grasshopper file (It's the code on top that matters).

Attachments:

• Test_Brugtranslatie.gh, 56 KB

Permalink Reply by martyn hogg on May 26, 2015 at 6:50am

You have used the Tangent vector of the curve to align the planes and I think this can flip depending on the curvature. Someone will probably be able to give you a better explanation.

Use a unit X vector...

Attachments:

• Test_Brugtranslatie.gh, 41 KB

Permalink Reply by Stijn Rosseel on May 27, 2015 at 2:00pm

Thx, it's not quitte what I wanted, but the alligned planes gave me an idea so now I worked out a solution that works!

The problem in your file is that the planes are actually "straight" vertical (they stay perpendicular to the world-horizontal plane (XY)) instead of bending with the curve and staying perpendicular to the curve (blue corner below). I have no acces to rhino or grasshopper at the moment but I'll try and post my solution tomorrow.

Permalink Reply by Joseph Oster on May 27, 2015 at 12:26pm

This is an interesting subject...  Take a look at the attached files using a single "rail" and "section" curve.

The "Sweep, one rail, one section" is the simplest, but it exhibits torsion as the section curve is twisted.

The "Torsion Free!" code uses horizontal planes to build vertical planes that orient the section curve as expected - and produce a roadway without torsion too.

P.S.  Be sure to use 'Shaded', not 'Raytraced with Neon', to see the rail and section curves.

Attachments:

• sweep2015May27a.gh, 32 KB
• sc2015May27a.3dm, 406 KB

Permalink Reply by Stijn Rosseel on May 27, 2015 at 2:17pm

Thx for your effort! I'll take a look at your solution tomorrow, but the results look very good!

In a very simular way I also found a solution to the problem (also creating the "no tortion" planes by using the horizontal planes in some way).
It feels like there should (or might) be an easier solution but perhaps there isn't :p

I'll post my solution tomorrow because I don't have acces to rhino and grasshopper at the moment.

Permalink Reply by Joseph Oster on May 27, 2015 at 2:51pm

This code is better:

Attachments:

• sweep2015May27b.gh, 32 KB
• sc2015May27b.3dm, 561 KB

Permalink Reply by Stijn Rosseel on May 28, 2015 at 12:57pm

Thx for your help guys!

Here's another way to create the perpendicular planes without tortion,

it might be usefull for someone...

Attachments:

• Perpendicular planes_No tortion.gh, 19 KB

Permalink Reply by nikos tzar on May 28, 2015 at 5:13pm

Hello all,

creating perpframes and aligning them with z vector should work:

Attachments:

• Perpendicularplanes_Notortion-nt.gh, 28 KB

Permalink Reply by Joseph Oster on May 28, 2015 at 9:51pm

That is an impressively simple bit of code!  I am surprised that "Align" to "Z" removes only the roll (torsion) from the PerpFrame, leaving the pitch intact.  This is what Stijn wanted, though I wanted to remove both, and did in my earlier code.

I made a few changes to your version so that it uses the same rail/sweep idea I did before.  I found it necessary to rotate my "sweep curve" (in yellow) before passing it to your "Orient" step:

Attachments:

• Perpendicularplanes_Notortion_2015May28c.gh, 11 KB
• sc2015May28a.3dm, 216 KB