OpenSCAD

dxf

Classic

List

Threaded

29 messages Options

lar3ry

Re: dxf

On 28 Jul 2020 at 8:57, cbernhardt wrote:

> I use an older version of AutoCAD (R14) and I have found that OpenSCAD will
> accept DXF files with AutoCAD entities Line, Arc, Polyline, Circle, Ellipse,
> and Polygon (actually a Polyline). If the Polyline has arc segments they
> must be eploded (broken into line and arc segments) to import into OpenSCAD
> correctly. AutoCAD 3DFACEs ( like a 3D triangle in an STL file ) will not
> import and AutoCAD SPLINE entities will not import.

I often have trouble with various types of files. I have had fairly good luck
converting DXF files to SVG, either by exporting them from Inkscape or
converting them to SVG on one of the on-line file converters.

I like this one: https://convertio.co/dxf-svg/

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

cacb

Re: dxf

In reply to this post by tp3

On 29.07.2020 00:58, Torsten Paul wrote:
> On 29.07.20 00:43, cbernhardt wrote:
>> Here is a ZIP file with spline.dxf, spline.jpg, and spline.txt with
>> coordinates and explanation.
>
> Thanks, that might help clarifying some details. It's currently
> trying but not quite getting the correct result for this file.
>

Here is my effort today trying to implement SPLINE in dxfread and
converting it to OpenSCAD.

It sort of works, but splines are *far* from well defined in DXF, it is
not at all clear how the "control points" are supposed to help, and the
same goes for the "knot values". I used the "fit points" to get the best
result matching the green spline in the original drawing provided by
cbernhardt.

Also, getting the boundary conditions right is not easy (and again the
DXF definition is not clear) so I applied a little work-around based on
the "periodic" setting for the spline to get something that looks right.

See attached PNG and generated .scad file.

P.S: This feature is not available in yesterday's released version of
dxfread, I implemented the feature based on today's example.

Carsten Arnholm

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

spline_dxfread.png (153K) Download Attachment

spline.scad (2K) Download Attachment

acwest

Re: dxf

I would expect that if it has control points and knot points, the expected implementation is a cubic bezier, which is very well defined, and quite easy to implement in OpenScad

On Wed, 29 Jul 2020, 07:48 Carsten Arnholm, <[hidden email]> wrote:

On 29.07.2020 00:58, Torsten Paul wrote:
> On 29.07.20 00:43, cbernhardt wrote:
>> Here is a ZIP file with spline.dxf, spline.jpg, and spline.txt with
>> coordinates and explanation.
>
> Thanks, that might help clarifying some details. It's currently
> trying but not quite getting the correct result for this file.
>

Here is my effort today trying to implement SPLINE in dxfread and
converting it to OpenSCAD.

It sort of works, but splines are *far* from well defined in DXF, it is
not at all clear how the "control points" are supposed to help, and the
same goes for the "knot values". I used the "fit points" to get the best
result matching the green spline in the original drawing provided by
cbernhardt.

Also, getting the boundary conditions right is not easy (and again the
DXF definition is not clear) so I applied a little work-around based on
the "periodic" setting for the spline to get something that looks right.

See attached PNG and generated .scad file.

P.S: This feature is not available in yesterday's released version of
dxfread, I implemented the feature based on today's example.

Carsten Arnholm
_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

cacb

Re: dxf

On 29.07.2020 14:06, A. Craig West wrote:
> I would expect that if it has control points and knot points, the
> expected implementation is a cubic bezier, which is very well defined,
> and quite easy to implement in OpenScad

Then there is no problem, just go ahead and do it.

Carsten Arnholm

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

Tim V. Shaporev

Re: dxf

In reply to this post by acwest

If my memory serves me right, AutoCAD splines are arbitrary order NURBS
i.e. the curve order is specified and (in theory) may be bigger than 3.

On 7/29/2020 3:06 PM, A. Craig West wrote:

> I would expect that if it has control points and knot points, the
> expected implementation is a cubic bezier, which is very well defined,
> and quite easy to implement in OpenScad
>
> On Wed, 29 Jul 2020, 07:48 Carsten Arnholm, <[hidden email]
> <mailto:[hidden email]>> wrote:
>
> On 29.07.2020 00:58, Torsten Paul wrote:
> > On 29.07.20 00:43, cbernhardt wrote:
> >> Here is a ZIP file with spline.dxf, spline.jpg, and spline.txt with
> >> coordinates and explanation.
> >
> > Thanks, that might help clarifying some details. It's currently
> > trying but not quite getting the correct result for this file.
> >
>
> Here is my effort today trying to implement SPLINE in dxfread and
> converting it to OpenSCAD.
>
> It sort of works, but splines are *far* from well defined in DXF, it is
> not at all clear how the "control points" are supposed to help, and the
> same goes for the "knot values". I used the "fit points" to get the
> best
> result matching the green spline in the original drawing provided by
> cbernhardt.
>
> Also, getting the boundary conditions right is not easy (and again the
> DXF definition is not clear) so I applied a little work-around based on
> the "periodic" setting for the spline to get something that looks right.
>
> See attached PNG and generated .scad file.
>
> P.S: This feature is not available in yesterday's released version of
> dxfread, I implemented the feature based on today's example.
>
> Carsten Arnholm
> _______________________________________________
> OpenSCAD mailing list
> [hidden email] <mailto:[hidden email]>
> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
>
>
> _______________________________________________
> OpenSCAD mailing list
> [hidden email]
> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
>

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

cbernhardt

Re: dxf

In reply to this post by cacb

cacb wrote

> Here is my effort today trying to implement SPLINE in dxfread and
> converting it to OpenSCAD.
>
> It sort of works, but splines are *far* from well defined in DXF, it is
> not at all clear how the "control points" are supposed to help, and the
> same goes for the "knot values". I used the "fit points" to get the best
> result matching the green spline in the original drawing provided by
> cbernhardt.
>
> Also, getting the boundary conditions right is not easy (and again the
> DXF definition is not clear) so I applied a little work-around based on
> the "periodic" setting for the spline to get something that looks right.
>
> See attached PNG and generated .scad file.
>
> P.S: This feature is not available in yesterday's released version of
> dxfread, I implemented the feature based on today's example.
>
> Carsten Arnholm

Here is a comparison of your spline to the original in the spline.dxf file.

<http://forum.openscad.org/file/t1309/spline2.jpg>

--
Sent from: http://forum.openscad.org/

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

cacb

Re: dxf

On 29.07.2020 15:02, cbernhardt wrote:
> Here is a comparison of your spline to the original in the spline.dxf file.
>
> <http://forum.openscad.org/file/t1309/spline2.jpg>

The difference is probably due to my difficulty applying the boundary
conditions correctly, whatever the correct procedure is. Oh well, such
things happen when the definition is unclear.

Regards
Carsten Arholm

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

tp3

Re: dxf

FWIW, below is the raw data from the DXF. And here's one of the
references to the specification:

http://docs.autodesk.com/ACD/2011/ENU/filesDXF/WS1a9193826455f5ff18cb41610ec0a2e719-79e1.htm

Note the extensive description of the fields below the table...

I'm not sure if the suspicious info "Adobe Flash Player is
required to view related information." would reveal anything
in addition, but I doubt that as the 270 page PDF has exactly
the same table and nothing more.

ciao,
Torsten.

spline: layer = 0, degree = 3
flags:
CLOSED
PERIODIC
PLANAR
start tangent: (0.94160, 0.33673)
end tangent: (0.94160, 0.33673)
knots (16): 0.00 0.00 0.00 0.00 5.11 8.83 14.25 18.25 22.00 26.57 31.73 34.60 37.91 37.91 37.91 37.91
ctrlp (12):
(35.07592, 11.58298 | 1.00000)
(36.67850, 12.15608 | 1.00000)
(39.89868, 11.51130 | 1.00000)
(43.86774, 14.14291 | 1.00000)
(49.59186, 13.56638 | 1.00000)
(49.93316, 8.12415 | 1.00000)
(46.40645, 5.11687 | 1.00000)
(41.67137, 5.96252 | 1.00000)
(37.72738, 6.49387 | 1.00000)
(33.76019, 7.79691 | 1.00000)
(34.03418, 11.21044 | 1.00000)
(35.07592, 11.58298 | 1.00000)

fitp (10):
(35.07592, 11.58298 | 0.00000)
(40.14785, 12.17122 | 0.00000)
(43.63480, 13.48344 | 0.00000)
(48.93316, 12.35222 | 0.00000)
(49.29544, 8.37029 | 0.00000)
(46.53305, 5.83634 | 0.00000)
(41.95925, 5.88159 | 0.00000)
(36.88732, 6.83182 | 0.00000)
(34.44193, 8.32504 | 0.00000)
(35.07592, 11.58298 | 0.00000)

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

-- Torsten

RevarBat

Re: dxf

In reply to this post by Tim V. Shaporev

If splines in DXF have knots, then it’s definitely using the more complex NURBS type splines, as Bezier’s of any degree don’t need knots.

https://en.wikipedia.org/wiki/Non-uniform_rational_B-spline

One nice thing about NURBS is that you can exactly model circular arcs, with the right knot vector.

- Revar

On Jul 29, 2020, at 5:41 AM, Tim V. Shaporev <[hidden email]> wrote:

If my memory serves me right, AutoCAD splines are arbitrary order NURBS i.e. the curve order is specified and (in theory) may be bigger than 3.

On 7/29/2020 3:06 PM, A. Craig West wrote:
I would expect that if it has control points and knot points, the expected implementation is a cubic bezier, which is very well defined, and quite easy to implement in OpenScad
On Wed, 29 Jul 2020, 07:48 Carsten Arnholm, <[hidden email] <[hidden email]>> wrote:
   On 29.07.2020 00:58, Torsten Paul wrote:
    > On 29.07.20 00:43, cbernhardt wrote:
    >> Here is a ZIP file with spline.dxf, spline.jpg, and spline.txt with
    >> coordinates and explanation.
    >
    > Thanks, that might help clarifying some details. It's currently
    > trying but not quite getting the correct result for this file.
    >
   Here is my effort today trying to implement SPLINE in dxfread and
   converting it to OpenSCAD.
   It sort of works, but splines are *far* from well defined in DXF, it is
   not at all clear how the "control points" are supposed to help, and the
   same goes for the "knot values". I used the "fit points" to get the
   best
   result matching the green spline in the original drawing provided by
   cbernhardt.
   Also, getting the boundary conditions right is not easy (and again the
   DXF definition is not clear) so I applied a little work-around based on
   the "periodic" setting for the spline to get something that looks right.
   See attached PNG and generated .scad file.
   P.S: This feature is not available in yesterday's released version of
   dxfread, I implemented the feature based on today's example.
   Carsten Arnholm
   _______________________________________________
   OpenSCAD mailing list
   [hidden email] <[hidden email]>
   http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org