rotate([45,45,0])
hi
i dont like this instruction because it is not clear in the formalism wether
x or y are calculated first
see this example, where tube is any module :
//rotate([0,45,0]) rotate([45,0,0]) tube();
rotate([45,0,0]) rotate([0,45,0]) tube();
rotate([45,45,0]) tube();
now, i needed something to automatically put a z-oriented object (say, a
cylinder) to any [x,y,z] direction
if it can be usegul :
module pointervers(v)
{
//v=[x,y,z]
x=v[0];
y=v[1];
z=v[2];
rho=sqrt(xx+yy+zz);
r=sqrt(xx+y*y);
phi=acos(z/rho);
if (r!=0)
{theta=acos(x/r)*sign(asin(y/r));}
else // cas d'un vecteur v=[0,0,-1]
{theta=0;}
echo(theta);
for (i=[0,$children-1])
{
rotate([0,0,theta]) rotate([0,phi,0]) children([i]);
}
}
--
Sent from: http://forum.openscad.org/
i dont like this instruction because it is not clear in the formalism
wether
x or y are calculated first
That's one of the reasons I've made some helper modules:
xrot(angle)
yrot(angle)
zrot(angle)
Usually, I only rotate on one axis at the time, and it gives neater code. I
also have similar modules for xtran, ytran and ztran, just to get cleaner
code.
--
Sent from: http://forum.openscad.org/
now, i needed something to automatically put a z-oriented object (say, a
cylinder) to any [x,y,z] direction
if it can be usegul :
I don't like the rotation by Euler angles either. In general it seems
more useful something like you did. My version is a bit more general and
simpler:
module rotFromTo(di,do)
if( norm(di-do)==0 || norm(di)==0 || norm(do)==0 )
children();
else
mirror(do/norm(do)+di/norm(di)) mirror(di) children();
It applies the least angle rotation that brings vector di to vector do.
@ Ronaldo, it's so amazing to me that this can be done in such a concise way.
I think it'd be good especially for alignment 'cos no knowledge of angle is
required.
In the mean time, readers are welcome to check out my version, which allows
for rotating solid obj or a list points about any axis in the space:
function rotM(pq,a) =
( let(
uv = ( pq[0]-pq[1])/norm(pq[1]- pq[0]) // unit vector
, x=uv[0]
, y=uv[1]
, z=uv[2]
, c=cos(a)
, s=sin(a)
, t=1-c
)
[ [ t*(xx)+c, t(xy)-sz, t*(xz)+sy ]
, [ t*(xy)+sz, t*(yy)+c, t(yz)-sx ]
, [ t*(xz)-sy, t*(yz)+sx, t*(z*z)+c ]
]
);
module rotObj( pq, a )
{
translate( pq[1] )
multmatrix(m= rotM( pq,a) )
{
translate( -1pq[1] )
children();
}
}
function rotPt(pt,pq,a)= rotM(pq,a)(pt-pq[0]) + pq[0] ;
function rotPts(pts,pq,a)= [ for(p=pts) rotPt(p,pq,a) ] ;
$ Runsun Pan, PhD $ libs: scadx , doctest , faces ( git ), offline doc ( git ), runscad.py ( 2 , git ), editor of choice: CudaText ( OpenSCAD lexer ); $ Tips ; $ Snippets
--
Sent from: http://forum.openscad.org/
From runsun runsun@gmail.com:
@ Ronaldo, it's so amazing to me that this can be done in such a concise
way.
I think it'd be good especially for alignment 'cos no knowledge of angle is
required.
The principle behind is that any rotation may expressed by composing two
reflections (mirror). In this case, you don't need to compute angles to
find the reflections.
Besides the operator rotFromTo, I have a functional version of it that
computes the rotation matrix:
function TrotFromTo(di,do) =
norm(di-do)==0 || norm(di)==0 || norm(do)==0 ?
[[1,0,0],[0,1,0],[0,0,1]] :
Tmirror(do/norm(do)+di/norm(di)) * Tmirror(di);
As expected, it requires a function returning the mirror transformation
matrix:
function Tmirror(v) =
norm(v)==0 ?
[[1,0,0],[0,1,0],[0,0,1]] :
let(u = v/norm(v))
[ [1,0,0] - 2*u[0]*u, [0,1,0] - 2*u[1]*u, [0,0,1] - 2*u[2]*u ];
hi
i am working on your codes, but i enconter something that i dont understand :
if (3!=0) {a=2;echo(a);}
echo(a);
the second echo(a); returns "undef" as if the variable a was defined only inside the {if}
how shall i understand this ?
Vincent
On 5/18/2018 12:18 AM, Vincent Douce Mathoscope wrote:
hi
i am working on your codes, but i enconter something that i dont understand :
if (3!=0) {a=2;echo(a);}
echo(a);
the second echo(a); returns "undef" as if the variable a was defined only inside the {if}
how shall i understand this ?
This might well be the most common OpenSCAD question.
OpenSCAD variables... aren't variables. They're constants, sort of,
with the last assignment winning and with inner scopes able to override
values from outer scopes. You can't change them, or at least not
exactly, and you especially can't change them inside an "if".
x=1;
echo(x=x);
x=2;
echo(x=x);
produces
ECHO: x = 2
ECHO: x = 2
and
x=1;
echo(x=x);
if (true) {
x=2;
echo (x=x);
}
echo (x=x);
produces
ECHO: x = 1
ECHO: x = 2
ECHO: x = 1
If you need to set a value conditionally, you need to use the ? :
operator, e.g.
a = (3 != 0) ? 2 : 1;
@ Ronaldo, Thx for sharing. Your code, as usual, is concise and elegant. It's
also much easier to use than mine for rotations around the origin.
$ Runsun Pan, PhD $ libs: scadx , doctest , faces ( git ), offline doc ( git ), runscad.py ( 2 , git ), editor of choice: CudaText ( OpenSCAD lexer ); $ Tips ; $ Snippets
--
Sent from: http://forum.openscad.org/
@ vincent,
Your issue with the variable can be solved by changing this part of your
code,
if (r!=0)
{theta=acos(x/r)*sign(asin(y/r));}
else // cas d'un vecteur v=[0,0,-1]
{theta=0;}
to
theta= r!=0? acos(x/r)*sign(asin(y/r)):0;
Also, note that your code is not a one-step rotation as you described in
"put a z-oriented object (say, a
cylinder) to any [x,y,z] direction". There's an extra twist other than the
rotation from z to [x,y,z].
Suppose we want to rotate a cube from P=[0,0,4] to Q=[2,2,2]. This would
imply a rotation about axis [[-1,1,0],[0,0,0],[1,-1,0]] (red line below) by
an angle= 54.7356 :
http://forum.openscad.org/file/t602/180518_rotation_test_starting_cube.png
Both of Ranaldo's and my codes,
module C(){ translate([-2,0,0]) cube([2,1,3]); }
P= [0,0,4];
Q= [2,2,2];
color("green") rotFromTo( di=P, do=Q) C(); // Ronaldo's code
color("green") rotObj( axis=[[-1,1,0],[1,-1,0]], a= 54.7356 ) C(); //
runsun's code
result in the same outcome:
http://forum.openscad.org/file/t602/180518_rotation_test_Ronaldo_runsun.png
Your code:
color("red") pointervers(Q) C();
results in :
http://forum.openscad.org/file/t602/180518_rotation_test_vincent_mathoscope.png
I'm not sure if that's intended. The twist is hard to spot if the starting
obj is a cylinder aligning w/ the z-axis. IMO, the extra twist makes it
harder to use in the context of "put a z-oriented objec to any [x,y,z]
direction".
$ Runsun Pan, PhD $ libs: scadx , doctest , faces ( git ), offline doc ( git ), runscad.py ( 2 , git ), editor of choice: CudaText ( OpenSCAD lexer ); $ Tips ; $ Snippets
--
Sent from: http://forum.openscad.org/
Module pointervers(v) produces a rotation that brings [0,0,1] to v.
My module rotFromTo(di,do) produces a rotation that brings di to do.
The point is that there is an infinite set of rotations that bring a
vector to another. For instance,
rotFromTo(w,d0) rotFromTo(di,w)
is able to bring di to do with a different rotation for (almost) any
other vector w.
rotFromTo chooses one specific rotation among that set: the rotation with
the minimum angle. The minimum rotation is unique and well defined: its
axis is cross(di,do).
In general, module pointvers(v) does not compute the minimum rotation
(except in trivial cases). That is fine but it has a major drawback: it is
not continuous in v. I would expect that for v very near to [0,0,1],
pointvers(v)
be very near to the identity and that is not true. For v near to [0,0,1],
the internally computed angle phi is near to 0 but theta is not.