Re: Help! How do you CNC a solidworks spline??
Cliff
wrote:
>On Jul 17, 12:40 pm, mike <mike_newsgro...@yahoo.com> wrote:
>> On Jul 17, 12:04 pm, raam...@gmail.com wrote:
>>
>>
>>
>> > On Jul 17, 1:41 pm, mike <mike_newsgro...@yahoo.com> wrote:
>>
>> > > Hi guys,
>>
>> > > I have a file with a spline cut in it, it looks like a half of an
>> > > ellipse. I'm using a Prototrak. When I save a DXF and use it to
>> > > generate the toolpath, I get a series of lines instead of a smooth
>> > > curve when I cut the metal. Does anyone know how to get a smooth
>> > > curve like what is in the part file?
>>
>> > > Thanks for any assistance here!
>> > > -Mike
>>
>> > dxf doesn't recognize spline entities; your spline has to broken into
>> > approximate arcs, a smooth result being the product of manipulating
>> > your cad to match the shape of the spline. it sux, and it looks like
>> > hell but that's the reality of it.
>>
>> how do you do this? I have the original solidworks file. Is there a
>> program that automates doing this?
>>
>> Thanks,
>> Mike
>
>
>I use to do this all the time with AutoCRAP files and DP Technologies
>Esprit X. In DP Technologies Esprit X it was called the LCA function.
>It converted lots of tiny lines to arcs so you got a smoother cut and
>the wire EDM didn't jerk.
So if he has the SW file AND had Esprit you'd suggest first converting
to a DXF and THEN into more little lines & arcs?
And you never thought of exporting from AutoCAD as IGES & importing
it that way into Esprit?
>What CAM software do you have or is all you have the ability to lay
>down toolpath on a .dxf with the Prototrak?
SW can export a DXF file and he clearly has the Prototrak
DXF option so all you need to tell him is how to adjust
the settings/tolerances for spline to DXF conversion in SW
(which I assume must exist).
>
>Jon Banquer
>San Diego, CA
>http://jonbanquer.blogspot.com/
<Snicker>
--
Cliff
Cliff
wrote:
>On Jul 17, 5:13 pm, mike <mike_newsgro...@yahoo.com> wrote:
>> On Jul 17, 4:34 pm, jon_banquer <jon_banq...@yahoo.com> wrote:
>>
>>
>>
>> > On Jul 17, 12:40 pm, mike <mike_newsgro...@yahoo.com> wrote:
>>
>> > > On Jul 17, 12:04 pm, raam...@gmail.com wrote:
>>
>> > > > On Jul 17, 1:41 pm, mike <mike_newsgro...@yahoo.com> wrote:
>>
>> > > > > Hi guys,
>>
>> > > > > I have a file with a spline cut in it, it looks like a half of an
>> > > > > ellipse. I'm using a Prototrak. When I save a DXF and use it to
>> > > > > generate the toolpath, I get a series of lines instead of a smooth
>> > > > > curve when I cut the metal. Does anyone know how to get a smooth
>> > > > > curve like what is in the part file?
>>
>> > > > > Thanks for any assistance here!
>> > > > > -Mike
>>
>> > > > dxf doesn't recognize spline entities; your spline has to broken into
>> > > > approximate arcs, a smooth result being the product of manipulating
>> > > > your cad to match the shape of the spline. it sux, and it looks like
>> > > > hell but that's the reality of it.
>>
>> > > how do you do this? I have the original solidworks file. Is there a
>> > > program that automates doing this?
>>
>> > > Thanks,
>> > > Mike
>>
>> > I use to do this all the time with AutoCRAP files and DP Technologies
>> > Esprit X. In DP Technologies Esprit X it was called the LCA function.
>> > It converted lots of tiny lines to arcs so you got a smoother cut and
>> > the wire EDM didn't jerk.
>>
>> > What CAM software do you have or is all you have the ability to lay
>> > down toolpath on a .dxf with the Prototrak?
>>
>> > Jon Banquer
>> > San Diego, CAhttp://jonbanquer.blogspot.com/
>>
>> I found out mastercam can do it, so I'll go that route. Thanks for
>> the help!
>
>Most CAM products can.
CLUE: He was ALREADY getting his DXF file from SW to the
Prototrak. How else could he have stated what he did?
>Be careful that you don't simplify the spline too much as what you are
>doing is an interpolation using arcs to replace the lines. If you go
>too far your part might no longer be in tolerance.
IF he took the spline *as a spline* to a CAM program why
would he need to "simplify" it instead of just cutting it as desired
to the tolerances desired using the CAM software as intended?
>
>Jon Banquer
>San Diego, CA
>http://jonbanquer.blogspot.com/
You've never done any of this, eh?
--
Cliff
Dave
Some apps (SWX being one) export splines to DXF as line segments. Even
hypochondriacs get sick and even JB gets one right once in a while....
Do you actually use SWX?
On Jul 22, 5:01 am, Cliff <Clhupr...@aol.com> wrote:
> On Thu, 17 Jul 2008 17:35:03 -0700 (PDT), jon_banquer <jon_banq...@yahoo.com>
Cliff
>You're pissing on your own leg.....
You are confused.
*I* posted "CLUE: He was ALREADY getting his DXF file from SW"
>Some apps (SWX being one) export splines to DXF as line segments.
And he needs somewhat better control of the approximation IF he going
to go that route.
*I* also posted:
[
SW can export a DXF file and he clearly has the Prototrak
DXF option so all you need to tell him is how to adjust
the settings/tolerances for spline to DXF conversion in SW
(which I assume must exist).
]
Can you tell him (& jb) about these settings?
>Even
>hypochondriacs get sick and even JB gets one right once in a while....
JB posted:
[
>I use to do this all the time with AutoCRAP files and DP Technologies
>Esprit X. In DP Technologies Esprit X it was called the LCA function.
>It converted lots of tiny lines to arcs so you got a smoother cut and
>the wire EDM didn't jerk.
]
To wihich *I* replied:
[
So if he has the SW file AND had Esprit you'd suggest first converting
to a DXF and THEN into more little lines & arcs?
And you never thought of exporting from AutoCAD as IGES & importing
it that way into Esprit?
]
>Do you actually use SWX?
JB posted:
[
>Be careful that you don't simplify the spline too much as what you are
>doing is an interpolation using arcs to replace the lines. If you go
>too far your part might no longer be in tolerance.
]
To wihich *I* replied:
[
IF he took the spline *as a spline* to a CAM program why
would he need to "simplify" it instead of just cutting it as desired
to the tolerances desired using the CAM software as intended?
]
BTW, I think that the DXF output probably just "apprximates" by
segments that have endpoints on the original curve. If so this means
that the segments are on the "inside" of the local curvatures of
the curves; quite a bias effect compared to the original curve.
A CAM package applied to the original curve should be able to
maintain a tolerance band +/- of the nominal curve. This would far
better approximate the curve even if the same total tolerance could
also be applied to te DXF conversion.
HTH
--
Cliff
Cliff
>You're pissing on your own leg.....
You are confused.
*I* posted "CLUE: He was ALREADY getting his DXF file from SW"
>Some apps (SWX being one) export splines to DXF as line segments.
And he needs somewhat better control of the approximation IF he going
to go that route.
*I* also posted:
[
SW can export a DXF file and he clearly has the Prototrak
DXF option so all you need to tell him is how to adjust
the settings/tolerances for spline to DXF conversion in SW
(which I assume must exist).
]
Can you tell him (& jb) about these settings?
>Even
>hypochondriacs get sick and even JB gets one right once in a while....
JB posted:
[
>I use to do this all the time with AutoCRAP files and DP Technologies
>Esprit X. In DP Technologies Esprit X it was called the LCA function.
>It converted lots of tiny lines to arcs so you got a smoother cut and
>the wire EDM didn't jerk.
]
To wihich *I* replied:
[
So if he has the SW file AND had Esprit you'd suggest first converting
to a DXF and THEN into more little lines & arcs?
And you never thought of exporting from AutoCAD as IGES & importing
it that way into Esprit?
]
>Do you actually use SWX?
JB posted:
[
>Be careful that you don't simplify the spline too much as what you are
>doing is an interpolation using arcs to replace the lines. If you go
>too far your part might no longer be in tolerance.
]
To wihich *I* replied:
[
IF he took the spline *as a spline* to a CAM program why
would he need to "simplify" it instead of just cutting it as desired
to the tolerances desired using the CAM software as intended?
]
BTW, I think that the DXF output probably just "apprximates" by
segments that have endpoints on the original curve. If so this means
that the segments are on the "inside" of the local curvatures of
the curves; quite a bias effect compared to the original curve.
A CAM package applied to the original curve should be able to
maintain a tolerance band +/- of the nominal curve. This would far
better approximate the curve even if the same total tolerance could
also be applied to te DXF conversion.
>On Jul 22, 5:01 am, Cliff <Clhupr...@aol.com> wrote:
HTH
--
Cliff
Cliff
>[
> SW can export a DXF file and he clearly has the Prototrak
>DXF option so all you need to tell him is how to adjust
>the settings/tolerances for spline to DXF conversion in SW
>(which I assume must exist).
>]
BTW, The quick & dirty way (& pretty simple) would be to
scale the spline up by a factor of 25.4 in inch units SW then
create the DXF file of it & then run it in metric mode on the
Prototrak (which I think the Prototrak can do).
--
Cliff
madcadman
"Dave" <dher...@hotmail.com> wrote in message
news:4f745311-33e0-44dc...@59g2000hsb.googlegroups.com...
> You're pissing on your own leg.....
>
Dave
hardly confused. Fonts and splines do not translate to splines on
export. They are converted to short line segments. There are no
settings which can change this fact, at least not up to 2006 releases.
The most efficient means I found to achieve the desired results is to
trace the splines using a 3 point arc. Tedious, but using the DXF as
is would result in a spline with a circumference of 12" being
converted to hundreds of lines and that means hundreds of extra lines
of GCode and an unacceptable finish on the cut edge.
Since you didn't reply to my direct question about your experience
with SWX, I can only assume you have none, based on what you did
post.
On Jul 22, 11:39 pm, Cliff <Clhupr...@aol.com> wrote:
> On Tue, 22 Jul 2008 15:59:07 -0700 (PDT), Dave <dherb...@hotmail.com> wrote:
> >You're pissing on your own leg.....
>
> You are confused.
> *I* posted "CLUE: He was ALREADY getting his DXF file from SW"
>
> >Some apps (SWX being one) export splines to DXF as line segments.
>
> And he needs somewhat better control of the approximation IF he going
> to go that route.
> *I* also posted:
> [
> SW can export a DXF file and he clearly has the Prototrak
> DXF option so all you need to tell him is how to adjust
> the settings/tolerances for spline to DXF conversion in SW
> (which I assume must exist).
> ]
>
> Can you tell him (& jb) about these settings?
>
> >Do you actually use SWX?
>
> JB posted:
> [>Be careful that you don't simplify the spline too much as what you are
> >doing is an interpolation using arcs to replace the lines. If you go
> >too far your part might no longer be in tolerance.
>
> [
jon_banquer
> I've exported hundreds of DXF files containing splines and fonts, I'm
> hardly confused. Fonts and splines do not translate to splines on
> export. They are converted to short line segments. There are no
> settings which can change this fact, at least not up to 2006 releases.
> The most efficient means I found to achieve the desired results is to
> trace the splines using a 3 point arc. Tedious, but using the DXF as
> is would result in a spline with a circumference of 12" being
> converted to hundreds of lines and that means hundreds of extra lines
> of GCode and an unacceptable finish on the cut edge.
>
> Since you didn't reply to my direct question about your experience
> with SWX, I can only assume you have none, based on what you did
> post.
>
Cliff Huprich has never used SolidWorks, Mastercam or DP Esprit. The
vast majority of information Cliff Huprich posts about CAD and CAM is
dead wrong. Anyone believing anything Cliff Huprich posts at this
point deserves what they get.
Despite these FACTS and the one's you just pointed out in regards to
DXF output that Cliff Huprich got so badly wrong, Cliff Huprich is the
largest poster to alt.machines.cnc and comp.cad.solidworks.
Cliff
>I've exported hundreds of DXF files containing splines and fonts, I'm
>hardly confused. Fonts and splines do not translate to splines on
>export.
I did not say or imply that they did (with a DXF export option) so
you are confused.
Don't go all banquer on us <g>.
>They are converted to short line segments.
Hence the OP's problem: he's not getting enough of them
or short enough ones to well-approximate his curve for his
need.
Sorry if you did not understand this but not my fault. Perhaps
you let jb confuse you with is cluelessness?
That often happens to some it seems. He copies words
& words are rational, right?
>There are no
>settings which can change this fact,
There may/should be a setting that controls the curve fitting tolerance,
right?
>at least not up to 2006 releases.
The DXF spec supports splines, sort of.
But the Prototrak accept spline input AFAIK.
>The most efficient means I found to achieve the desired results is to
>trace the splines using a 3 point arc. Tedious, but using the DXF as
>is
The OP very clearly has the Prototrak DXF option.
>would result in a spline with a circumference of 12" being
>converted to hundreds of lines and that means hundreds of extra lines
>of GCode
If they are the right extra lines of code problem solved for the OP.
>and an unacceptable finish on the cut edge.
HUH?
>Since you didn't reply to my direct question about your experience
>with SWX, I can only assume you have none, based on what you did
>post.
And which you seem to have had a few problems with, sorry.
This stuff is just basic CAD applications nuts & boltz.
HTH
--
Cliff
Cliff
wrote:
>On Jul 23, 11:51 am, Dave <dherb...@hotmail.com> wrote:
>> I've exported hundreds of DXF files containing splines and fonts, I'm
>> hardly confused. Fonts and splines do not translate to splines on
>> export. They are converted to short line segments. There are no
>> settings which can change this fact, at least not up to 2006 releases.
>> The most efficient means I found to achieve the desired results is to
>> trace the splines using a 3 point arc. Tedious, but using the DXF as
>> is would result in a spline with a circumference of 12" being
>> converted to hundreds of lines and that means hundreds of extra lines
>> of GCode and an unacceptable finish on the cut edge.
>>
>> Since you didn't reply to my direct question about your experience
>> with SWX, I can only assume you have none, based on what you did
>> post.
>>
>
>Cliff Huprich has never used SolidWorks, Mastercam or DP Esprit. The
>vast majority of information Cliff Huprich posts about CAD and CAM is
>dead wrong. Anyone believing anything Cliff Huprich posts at this
>point deserves what they get.
Proof yet again that you are a drooling idiot.
>Despite these FACTS and the one's you just pointed out in regards to
>DXF output that Cliff Huprich got so badly wrong,
Did you grasp one word in ten?
I doubt it.
>Cliff Huprich is the
>largest poster to alt.machines.cnc and comp.cad.solidworks.
<Snicker>
So where did you get your confused stuff from this time?
>Jon Banquer
>San Diego, CA
There is a jon banquer in a low-income assisted
housing place (or nursing home?) or something there I gather.
--
Cliff
castleb...@att.net
"Cliff" <Clhu...@aol.com> wrote in message
news:8ngf8416v8mcpdj5h...@4ax.com...
Splines in a DXF file are called splines with points on the curve given. I
would suggest that the cam software does not understand the concept of
splines.
Bob
Cliff
Other data is needed too.
http://www.autodesk.com/techpubs/autocad/acad2000/dxf/spline_dxf_06.htm
>I would suggest that the cam software does not understand the concept of
>splines.
Are all NURBS curves splines? Not in this sense I think.
Does the SW DXF export function export such as spline approximations
of some sort (given the proper settings)? Dave said not but .... AutoCad
seems to be (from a brief search) able to import splines from SW via
DXF.
What of the Prototrak DXF option? I found no mention of any spline
capability for the Prototrak.
>Bob
--
Cliff
Cliff
Added back xposting groups.
>> raa...@gmail.com wrote:
>> dxf doesn't recognize spline entities
>------------------------------------------
>
>[begin quote]
>
>DXF Group Codes
>(version number u14.1.03.)
>
>--------------------
> ENTITIES Section
>
> SPLINE
>--------------------
>
>
>The following group codes apply to spline entities.
>
>Spline group codes Group codes Description
>100
> Subclass marker (AcDbSpline)
>
>210
> Normal vector (omitted if the spline is nonplanar)
>DXF: X value; APP: 3D vector
>
>220, 230
> DXF: Y and Z values of normal vector
>
>70
> Spline flag (bit coded):
>1 = Closed spline
>2 = Periodic spline
>4 = Rational spline
>8 = Planar
>16 = Linear (planar bit is also set)
>
>71
> Degree of the spline curve
>
>72
> Number of knots
>
>73
> Number of control points
>
>74
> Number of fit points (if any)
>
>42
> Knot tolerance (default = 0.0000001)
>
>43
> Control-point tolerance (default = 0.0000001)
>
>44
> Fit tolerance (default = 0.0000000001)
>
>12
> Start tangent--may be omitted (in WCS).
>DXF: X value; APP: 3D point.
>
>22, 32
> DXF: Y and Z values of start tangent--may be omitted (in WCS).
>
>13
> End tangent--may be omitted (in WCS).
>DXF: X value; APP: 3D point.
>
>23, 33
> DXF: Y and Z values of end tangent--may be omitted (in WCS)
>
>40
> Knot value (one entry per knot)
>
>41
> Weight (if not 1); with multiple group pairs, are present if all are not 1
>
>10
> Control points (in WCS) one entry per control point.
>DXF: X value; APP: 3D point
>
>20, 30
> DXF: Y and Z values of control points (in WCS) (one entry per control point)
>
>11
> Fit points (in WCS) one entry per fit point.
>DXF: X value; APP: 3D point
>
>21, 31
> DXF: Y and Z values of fit points (in WCS) (one entry per fit point)
>
>[end quote]
>
>http://usa.autodesk.com/adsk/servlet/usearch/results?la=en&nh=10&col=wacjprd&col=usuppprd&siteID=123112&catID=123155&id=2088334&qt=d
>xf+reference
>
>http://usa.autodesk.com/adsk/servlet/ps/item?siteID=123112&id=2882295&linkID=9240617
>
>HTH
>
>Bonus question:
>What's the difference between a Solidworks spline and a Solid Edge spline?
>
>
castleb...@att.net
0
SECTION
2
ENTITIES
0
SPLINE
5
17A
330
23
100
AcDbEntity
8
VISABLE
100
AcDbSpline
210
0.0
220
0.0
230
1.0
70
8
71
3
72
12
73
8
74
6
42
0.0000000001
43
0.0000000001
44
0.0000000001
13
0.983148089309517
23
0.182810925513346
33
0.0
40
0.0
40
0.0
40
0.0
40
0.0
40
0.5061414898570419
40
0.9199702471070123
40
1.575345136293664
40
2.045039716453304
40
2.769260067854284
40
2.769260067854284
40
2.769260067854284
40
2.769260067854284
10
13.43328730535758
20
16.85005512985181
30
0.0
10
13.58842145696185
20
16.78487631740688
30
0.0
10
13.87039558207541
20
16.66640634528891
30
0.0
10
14.34456513808076
20
16.42415660731785
30
0.0
10
14.89601632531026
20
16.49226158568692
30
0.0
10
15.40840974183308
20
16.92516579383252
30
0.0
10
15.83519929917696
20
16.65671426629983
30
0.0
10
16.07253791741661
20
16.70084606387157
30
0.0
11
13.43328730535758
21
16.85005512985181
31
0.0
11
13.89715557325682
21
16.64755713531641
31
0.0
11
14.28104652301536
21
16.49301918198279
31
0.0
11
14.93152857018393
21
16.57295260232628
31
0.0
11
15.35274227579633
21
16.78077948421505
31
0.0
11
16.07253791741661
21
16.70084606387157
31
0.0
0
ENDSEC
Cliff
>On Wed, 23 Jul 2008 16:49:51 -0600, "nitnit" <n...@no.mail> wrote:
>
> Added back xposting groups.
>
>>> raa...@gmail.com wrote:
>>> dxf doesn't recognize spline entities
>>Bonus question:
>>What's the difference between a Solidworks spline and a Solid Edge spline?
Let's wait & see what jb sez. IF he wants to claim a clue,
which I doubt.
Bound to be funny ....
--
Cliff
castleb...@att.net
http://en.wikipedia.org/wiki/Spline_interpolation
My conclusion, if your cam program uses the same interpolation method as
your CAD program you get the expected results, If not?
Bob
Cliff
A huge IF.
They are not the same in any case. A decent CAD/CAM or CAM
application will interpolate along a spline (or many other curves/surfaces)
within a tolerance limit. A tolerance outside (leaving some extra stock)
and one inside (cutting some extra stock away) is typical. Either tolerance
might be set to zero but never both ad it's pretty typical to use the
same tolerance for both. The tighter the total tolerance band the more
interpolation points that are generated to approximate te curve.
Only lines (or flat surfaces) and arcs/circles can have exact representations
in most CNC machine G-code (which only has motion words & language/control
for linear & circular moves).
Though some controls do have/allow spline interpolation functions/code
which require somewhat special postprocessors & CAM applications.
Nor is a NURBS curve guaranteed to be a spline <G>.
Nor does the Prototrak seem to have any spline interpolation
functions AFAIK.
Too bad that clueless jb knows none of this.
>Bob
>
--
Cliff
Cliff
>
>"Cliff" <Clhu...@aol.com> wrote in message
>news:ge0e84dna7d3v6hif...@4ax.com...
>> On Wed, 23 Jul 2008 05:36:25 -0400, Cliff <Clhu...@aol.com> wrote:
>>
>>>[
>>> SW can export a DXF file and he clearly has the Prototrak
>>>DXF option so all you need to tell him is how to adjust
>>>the settings/tolerances for spline to DXF conversion in SW
>>>(which I assume must exist).
>>>]
>>
>> BTW, The quick & dirty way (& pretty simple) would be to
>> scale the spline up by a factor of 25.4 in inch units SW then
>> create the DXF file of it & then run it in metric mode on the
>> Prototrak (which I think the Prototrak can do).
>> --
>> Cliff
>
>http://en.wikipedia.org/wiki/Spline_interpolation
NURBS: http://en.wikipedia.org/wiki/Nonuniform_rational_B-spline
nitnit
>> method as your CAD program you get the expected results, If not?
> A huge IF.
> They are not the same in any case. A decent CAD/CAM or CAM
> application will interpolate along a spline
Is there some confusion here? It looks, to me, like on one hand
someone's talking about interpolation and on the other someone's
talking about the approximation of, for instance, degree three
curves with either degree 1 or degree 2 curves or their analytic
line or arc equivalents.
> Nor is a NURBS curve guaranteed to be a spline <G>.
When is a Non-Uniform Rational B-Spline not a "spline"?
Is it meant that not all "splines" can be precisely represented by NURBS?
Are there any in common CAx usage today or, even, since the '50s or '60s?
Cliff
>>> My conclusion, if your cam program uses the same interpolation
>>> method as your CAD program you get the expected results, If not?
>
>> A huge IF.
>> They are not the same in any case. A decent CAD/CAM or CAM
>> application will interpolate along a spline
>
>
>Is there some confusion here?
Not on my part.
>It looks, to me, like on one hand
>someone's talking about interpolation
Even there there might be multiple subjects, such
as precise interpolation *of the curve* or the methods
by which such forms might be actually made by CNC/CAM
software/hardware. And how to get from NURBS curve
to part..
>and on the other someone's
>talking about the approximation of, for instance, degree three
>curves with either degree 1 or degree 2 curves or their analytic
>line or arc equivalents.
Someone mentioned the later but that's not really related to
the subject.
>> Nor is a NURBS curve guaranteed to be a spline <G>.
>
>
>When is a Non-Uniform Rational B-Spline not a "spline"?
Simple splines are not, in general, rational unless the denominator is
1 & you count that.
>Is it meant that not all "splines" can be precisely represented by NURBS?
IIRC all splines can be but not the converse.
>Are there any in common CAx usage today or, even, since the '50s or '60s?
Yes. NURBS is fairly new and possible mostly due to
increased computing speed & power (and probably some fancy
polynomial approximation computational math breakthroughs).
Some years ago cubic (max) rationals on super-minis were state of the art
(Example: the solids modeler that was being developed for CALMA).
BTW, There's some good stuff in the back of the IGES spec.
--
Cliff
nitnit
> Someone mentioned the later but that's
> not really related to the subject.
I see. My bad.
>>> Nor is a NURBS curve guaranteed to be a spline <G>.
>> When is a Non-Uniform Rational B-Spline not a "spline"?
> Simple splines are not, in general, rational
> unless the denominator is 1 & you count that.
Yes. I do. Relatively few programs will, excluding rational second
degree conic representations, create a rational Bezier or B-Spline.
So when we're talking about NURBS we're usually talking about the
(W = 1) nonrational forms.
But the statement was:
"Nor is a NURBS curve guaranteed to be a spline.".
A NURBS curve may not be rational, it may not even wiggle, but it is
always a spline.
>> Is it meant that not all "splines" can be precisely represented by NURBS?
> IIRC all splines can be but not the converse.
>> Are there any in common CAx usage today or, even, since the '50s or '60s?
> Yes.
The question was: Are there any splines [that cannot be precisely represented
by NURBS] in common usage today? I should hope there are not.
> NURBS is fairly new
'Fairly new' is fairly old by 'CAD for the common man' standards.
1995 - "AutoSurf version 2.1 for use with AutoCAD 13, integrates
precise NURBS technology and Advanced Surfacing."
The MES (Solutions 3000) code was acquired in '92 making it, roughly,
twenty years old.
> BTW, There's some good stuff in the back of the IGES spec.
Yes! I've seen that.
I even understand the occasional bit and piece of it. ;^)
Cliff
>
>> Someone mentioned the later but that's
>> not really related to the subject.
>
>I see. My bad.
>
>
>>>> Nor is a NURBS curve guaranteed to be a spline <G>.
>>> When is a Non-Uniform Rational B-Spline not a "spline"?
>> Simple splines are not, in general, rational
>> unless the denominator is 1 & you count that.
>
>Yes. I do.
That's about like saying that all rational numbers are integers.
"Simple splines are not, in general, rational unless the denominator is 1
& you count that."
>Relatively few programs will, excluding rational second
>degree conic representations, create a rational Bezier or B-Spline.
>So when we're talking about NURBS we're usually talking about the
>(W = 1) nonrational forms.
Not AFAIK. Were that the case you'd not need to have NURBS.
>But the statement was:
>
> "Nor is a NURBS curve guaranteed to be a spline.".
>
>A NURBS curve may not be rational, it may not even wiggle, but it is
>always a spline.
IOW a denominator of 1. Not.
>>> Is it meant that not all "splines" can be precisely represented by NURBS?
>> IIRC all splines can be but not the converse.
>>> Are there any in common CAx usage today or, even, since the '50s or '60s?
>> Yes.
>
>The question was: Are there any splines [that cannot be precisely represented
>by NURBS] in common usage today? I should hope there are not.
Wilson- Fowler splines?
>> NURBS is fairly new
>
>'Fairly new' is fairly old by 'CAD for the common man' standards.
>
> 1995 - "AutoSurf version 2.1 for use with AutoCAD 13, integrates
> precise NURBS technology and Advanced Surfacing."
ComputerVision had NURBS up to degree 17 by no later than 1988.
DelCAM/DUCT may have been even earlier.
>The MES (Solutions 3000) code was acquired in '92 making it, roughly,
>twenty years old.
>
>
>> BTW, There's some good stuff in the back of the IGES spec.
>
>Yes! I've seen that.
>I even understand the occasional bit and piece of it. ;^)
Which pieces?
Try explaining any of it to poor jb <G>.
I have some of the books mentioned (Faux & Pratt as
an example) and have done odd things with IGES (which
the CAD system directly would have taken a lot of time
to do manually, etc.)
--
Cliff
nitnit
Is it?
>> Relatively few programs will, excluding rational second
>> degree conic representations, create a rational Bezier or B-Spline.
>> So when we're talking about NURBS we're usually talking about the
>> (W = 1) nonrational forms.
> Not AFAIK.
If you know of a (we're "usually" talking) main stream 'MCAD' application
that will create a rational degree 3, or higher, Bezier or B-Spline do me
the favor of pointing it out.
> Were that the case you'd not need to have NURBS.
Non-uniform knots? Compatibility? It's probably safe to say there are far
more programs capable of importing rational entities without screwing them
up than there are creating them, not unlike higher order entities.
>> but it is always a spline.
> IOW a denominator of 1. Not.
"Not" as in; not a rational spline?
> Wilson-Fowler splines?
In "common CAx usage today"?
> ComputerVision had NURBS up to degree 17 by no later than 1988.
> DelCAM/DUCT may have been even earlier.
How do you define "fairly new" then?
> Which pieces?
Appendix K: Glossary
Spline: A piecewise continuous polynomial.
Cliff
>> That's about like saying that all rational numbers are integers.
>
>Is it?
We all know that 1+1=3 for sufficiently large 1, right?
>>> Relatively few programs will, excluding rational second
>>> degree conic representations, create a rational Bezier or B-Spline.
>>> So when we're talking about NURBS we're usually talking about the
>>> (W = 1) nonrational forms.
>
>> Not AFAIK.
>
>If you know of a (we're "usually" talking) main stream 'MCAD' application
>that will create a rational degree 3, or higher, Bezier or B-Spline do me
>the favor of pointing it out.
Just for starters Computervision & Unigraphics can both
create NURBS curves & surfaces of degrees higher than 3.
(Aside; This certainly implies that ParaSolid supports such as well
(though I doubt a vendor using it is required to do so).)
What was once intended for CALMA solids was limited to cubics
by design & intent.
>> Were that the case you'd not need to have NURBS.
>
>Non-uniform knots? Compatibility? It's probably safe to say there are far
>more programs capable of importing rational entities without screwing them
>up than there are creating them, not unlike higher order entities.
NURBS without the R would be the ever-popular NUBS.
Just ask jb.
>>> but it is always a spline.
>
>> IOW a denominator of 1. Not.
>
>"Not" as in; not a rational spline?
>
>
>> Wilson-Fowler splines?
>
>In "common CAx usage today"?
Such as by/in Catia or some of the "latest & greatest" stuff jb
has come up with as "THE Wave of the Future!" in years past?
>> ComputerVision had NURBS up to degree 17 by no later than 1988.
>> DelCAM/DUCT may have been even earlier.
>
>How do you define "fairly new" then?
Umm ... how new are you at this?
>> Which pieces?
>
>Appendix K: Glossary
> Spline: A piecewise continuous polynomial.
Rational & non-uniform <g>?
--
Cliff
nitnit
> We all know that 1+1=3 for sufficiently large 1, right?
We all know that has nothing to do with the four coordinate values
describing a control vertex as applied to common CAx use NURBS
application, right?
>> main stream 'MCAD' application that will create a rational degree 3
> Just for starters Computervision & Unigraphics can both
The question stands.
Computervision "can"?
Yes; CADDS, UG, Catia can.
Rhino, Alias and, I assume, ICEM can.
Alibre?
Cobalt?
Inventor?
Pro/Engineer?
Solid Edge?
Solidworks?
Any other 'main stream MCAD' programs?
Or modeling packages associated with CAM software?
It would be interesting information and we both stand to learn something.
> (Aside; This certainly implies that ParaSolid supports such as well
> though I doubt a vendor using it is required to do so).)
You keep stumbling over that word "support", keep assuming it means more
than it does.
> What was once intended for CALMA solids was
> limited to cubics by design & intent.
As opposed to; by accident?
Pro/E "supports" higher order NURBS; e.g. will import and interact with
them. It can / will export them intact; e.g. without altering their
definition. By current "design and intent" (I suppose, as PTC bought
Computervision / CADDS almost ten years ago) it will not create an entity
with degree greater than 3 (cubic) nor will it create a rational degree 3
entity.
> NURBS without the R would be the ever-popular NUBS.
And without NU that would be the ever-so-popular BS?
"The generalization of B-spline curves to NURBS (Nonuniform Rational
B-splineS) has become the standard curve and surface form in the
CAD/CAM industry. They offer A UNIFIED REPRESENTATION of spline and
conic geometries: every conic as well as every spline allows a
piecewise rational polynomial representation. The origin of the
term NURBS is unclear; but the TERM was certainly a bad choice:
it explicitly excludes the popular uniform B-spline curves."
- Farin; "A History of Curves and Surfaces in CAGD"
And without B you have S which is all that is inferred by the term NURBS;
it is going to be a spline.
> Umm ... how new are you at this?
Not so new that I think this tap dancing is much more interesting than,
or anticipate it being more informative than "I Love Lucy" reruns or
contemplating the nature and meaning of a "kernel".
Cliff
>
>> We all know that 1+1=3 for sufficiently large 1, right?
>
>We all know that has nothing to do with the four coordinate values
>describing a control vertex as applied to common CAx use NURBS
>application, right?
But it often gets BottleBob going <G>.
>>> main stream 'MCAD' application that will create a rational degree 3
>> Just for starters Computervision & Unigraphics can both
>
>The question stands.
>Computervision "can"?
>Yes; CADDS, UG, Catia can.
>Rhino, Alias and, I assume, ICEM can.
>Alibre?
>Cobalt?
>Inventor?
>Pro/Engineer?
>Solid Edge?
>Solidworks?
>Any other 'main stream MCAD' programs?
>Or modeling packages associated with CAM software?
It would do little good for a CAD/CAM package to allow
the creation of a surface which it cannot machine though
some used to allow the users to add their own "foreign"
types (not really sure how that worked, exactly).
They had to supply their own evaluation software too IIRC.
>It would be interesting information and we both stand to learn something.
Probably.
>> (Aside; This certainly implies that ParaSolid supports such as well
>> though I doubt a vendor using it is required to do so).)
>
>You keep stumbling over that word "support", keep assuming it means more
>than it does.
??
If UG can use it in UG the ParaSolid kernel can support it.
That does not mean that all other systems that use ParaSolid
also do so AFAIK. They have their own applications software
and could well choose not to allow more than cubics, etc. though
I have no idea what would happen in they do that and then import
a higher degree PasaSolid file.
>> What was once intended for CALMA solids was
>> limited to cubics by design & intent.
>
>As opposed to; by accident?
No; they thought (and considering compute power & cost
of the day) that cubic rationals could support most (if not all)
expected needs. More segments & patches?
>Pro/E "supports" higher order NURBS; e.g. will import and interact with
>them. It can / will export them intact; e.g. without altering their
>definition.
That's always a crucial test (and a good one).
>By current "design and intent" (I suppose, as PTC bought
>Computervision / CADDS almost ten years ago) it will not create an entity
>with degree greater than 3 (cubic) nor will it create a rational degree 3
>entity.
That would be Pro-E I expect, not CADDS.
IIRC They have migrated some of the CAM stuff from CADDS to Pro-E
but my impression (which is a bit old & stale) is that they only got
so far.
>> NURBS without the R would be the ever-popular NUBS.
>
>And without NU that would be the ever-so-popular BS?
Or banquercadcam: http://www.geocities.com/banquercadcam/
> "The generalization of B-spline curves to NURBS (Nonuniform Rational
> B-splineS) has become the standard curve and surface form in the
> CAD/CAM industry. They offer A UNIFIED REPRESENTATION of spline and
> conic geometries: every conic as well as every spline allows a
> piecewise rational polynomial representation. The origin of the
> term NURBS is unclear; but the TERM was certainly a bad choice:
> it explicitly excludes the popular uniform B-spline curves."
> - Farin; "A History of Curves and Surfaces in CAGD"
>
>
>And without B you have S which is all that is inferred by the term NURBS;
>it is going to be a spline.
Which we had before. From the early days of APT IIRC.
>> Umm ... how new are you at this?
>
>Not so new that I think this tap dancing is much more interesting than,
>or anticipate it being more informative than "I Love Lucy" reruns or
>contemplating the nature and meaning of a "kernel".
Quite.
--
Cliff