Some surfaces behaviour makes me confused

[design-engine]

isoparms are different than isplines. PTC uses isolines as the tangency edge where two surfaces join. Isoparms are the lines in between the surface patches.

There is one isoparm in this surface because the two adjacent
curves have one extra span. hence the 3 degree 2 span curve
Notice the one darker line...

no isoparms just patch precision


Edited by: design-engine

 

[design-engine]

Helio Castaoneves Bell helmet presented
in multiple formats. The first Pro/E shaded mode where you cant see
tangency lines simply because it's shaded. The second is the Alias
model. It shows tangency lines and isoparms. The third image shows
Pro/E wire mode. No isoparms but it does show tangency lines.



In my g2 wrokshop i tell participants: "If you want to sit at the cool table for lunch at GM you can't have isoparms in your geometry"


Edited by: design-engine

 

[jeff4136]

Bart, I can't tell exactly what I'm looking at from pics.
My guess, though, is that the "darker line" is an isoparametric
curve at a knot location. The rest are just uniformly spaced
isoparametric curves within the knot span. (Open a Rhino file
and adjust isoparm density to see something similar, only there
is no difference in color for the curves at knot locations.)


What you are calling "isolines" or "tangency edges" are simply
surface boundary edges within a quilt.


"Patch precision" is an undefined term. Unless we break a NURBS
entity down into it's basis spline elements I don't believe it
has any meaning and, even then, I believe it will simply be another
term for "knot span".


I don't think I'll ever be given a place at the cool guy lunch table,
anyway.

 

[design-engine]

isolines - ispmetric lines within a surface patch that dictate surface complexity. For example in Pro/ENGINEER when you model a spline with two end points (no internal point) you get a 3 degree curve y=x^3. When you add the extra point you get two y=x^3 stuck together. Those two those two jointed together creates an isoparm in the surface.

isolines are the tangency line edge that comes from a curve with two separate spline curves joined together in a sketch. The sruface results in an isoline. In the top image above with the two three part boundary quarters.... the top example has one isparm in the surface. Pro/ENGINEER does not display those... yet.

there is no definition for ispoline in proe... yet.

patchprecision= is the same as mesh in Pro/E where you can make the uv lines anythign you want.

It is so much easier to just show it .... maybe I create a video this weekend. I am geared up to do those youtube videos.

I am going to the gym to stare at somthing other than isoparms




Edited by: design-engine

 

[jeff4136]

> For example in Pro/ENGINEER when you model a spline with two
> end points (no internal point) you get a 3 degree curve y=x^3.


That's correct.
It's worth noting that the result is a degree 3 Bezier curve.


> When you add the extra point ...


A "knot".


... you get two y=x^3 stuck together.


Basis splines / functions.
Create that same curve in Rhino and ConvertToBeziers. Turn on
CV's and you'll see two degree 3 Bezier curves each with 4 CVs.
("Analogous to B

 

[muadib3d]

I struggled quite a bit with round case shown below. Pro\e gave up at start. I wasn`t so soft.

and at last





not the best but worth the time

Hey, guys what are yours workarounds for following case?

I tried to fit this new solution into mentioned above case. However something is wrong wit my boundary conditions and result are "strange" to say at least





the rest of Your both discuss I have to read twice to get a clue))))

 

[design-engine]

be careful counting CV's in Rhino or Alias...

3degree 2 span counts four CV's which is what we get in Pro/E and SW.ve


What i want in Pro/ENGINEER is up to 5 and 7 degree curve with one span so we get full control of the CV's hence more control of a curve.

Thats what I have been pushing for for ten years at least. ;( I am getting pushy and will in my talk show with example why.

What IGES export you want Jeff?

 

[dgs]

I haven't been following this, but I just read through the last 2-3 pages. Now I feel like a Pro|E surfacing knuckledragger.

Bart - You andI have to talk about this stuff in Long Beach. You are coming to a RUG in Ohio sometime? Is that NOPUG inthe Cleveland area? Their web site seems to be down.

 

[design-engine]

I think the consultants in the OHIO Rug don't like me to speak there for many reasons.... Maybe I'll leave a message each day Ill get a call back. I speak at the Milwaukee badgerland Pro/USER conference on a regular basis. My presentation does not sell anything (except alternate training to PTC style book training or PTC UNIVERSITY) which is probably the biggest problem. And since design-engine is considered a service provider and not a technical presentation ... my presentation tends to get lumped into a sales presentation category which is not desierable at the conferences.... until they see my presentation.

Put in a good word for me to speak at that Ohio one and Ill drive out. You got to get me a beer tho.


Edited by: design-engine

 

[muadib3d]

jeff could you explain me what is behind this sequence of values used in your relation?



I understand what trajpar is, I know what is the dimension d35. What I don`t know though, is the reason of use value 55? I understand that the whole relations must keep the same angle as in VSS before, to meet tangency at the connection point between these two surfaces

 

[muadib3d]

I struggled quite a bit with round case shown below. Pro\e gave up at start. I wasn`t so soft.

Hey, guys what are yours workarounds for following case?

I tried to fit this new solution into mentioned above case. However something is wrong wit my boundary conditions and result are "strange" to say at least





the rest of Your both discuss I have to read twice to get a clue))))

I`ve solved it. IMHO results are not such good as in my own approach, but jeff thx anyway. I own You a bear.

The point is there was no problem with conditions, only in BB settings. After I turned of "Add inner edge tangency" problem disappeared.





being so close, and at same time being so far though.

 

[dgs]

Put in a good word for me to speak at that Ohio one and Ill drive out. You got to get me a beer tho.

I thought you had something set up already.


I'm actually not part of the Northern OhioRUG, although I've considered going up to theirmeetings. It's about 2 hours away from here. The local Columbus RUG is dead. It hasn't been active in 4-5 years.

 

[jeff4136]

Bart,


> What IGES export you want Jeff?


I was interested in seeing the surfs shown in pic on p14 this discussion
("This is the classic g2 workshop in Alias.") Just the blend and bounding
surfs but make it easy on yourself. I wouldn't mind seeing the whole thing.


> be careful counting CV's in Rhino


Unless the exporting program caps (i.e. intf_out_max_bspl_degree) or
otherwise alters exported NURBS entities as a result of some configuration
setting the CVs, knot vectors, etc. (e.g. entities that define the curve
or surface) should not be changed as a result of translation. The only
difference between what I see in Rhino and what you see in Alias, for
example, will be differences in the way an individual program choses to
display the entities (i.e. the darker knot location isoparm).


> What i want in Pro/ENGINEER is up to 5 and 7 degree curve


I'd like to see that as well. Degree 5 Beziers would simplify curvature
matching and even low end programs create degree 5 G2 blends. I wonder
though; where's the profitability for PTC? Being truly competitive in the
Class A arena is the deep end of geometry engine development, there'd be a
tremendous amount of code to rewrite as a result and big money markets have
a lot of inertia. If they think the issue significant in maintaining their
comfortable market position, e.g. not losing share to SW over related
issues, they may make the investment. I'm not so sure it is that
significant. Higher degree entities help but they are not magic button
solutions that will transform Class A apprentices into journeymen, higher
order entities without higher order continuity functions improve little and
I can't imagine PTC ~really~ caring if they have a favored place among
Class A elitists.
... That's my erstwhile thoughts and I ~am~ a lot more interested in
delving into what I ~can~ do with what I have at my disposal than lamenting
over what I don't have. Well, I do spend some time wishing for VSS surf
edge traj curvature references and degree 5 Bezier section curves and ... ;^)


- - - - - -


Jacek,


> sd4 = 55 + (d35 - 55) ...


"d35 - d55" rotates (twists) the section.
"cos(90 * trajpar)" adjusts the rate.
2008-02-21_132825_twist-linear-nonlinear--wf2--.prt.zip

 

[muadib3d]

I own You a beer, not a bear, Jeff. Or just in polish - wisze ci piwo)

Anyway I gonna try to set these surfaces with splines sketches, maintaining the curvature at a connection point between them.



I `ve made already some efforts, however the devil sticks in details, and I do not know how to set up relations right though.

 

[muadib3d]

I`ve hope I am not introducing a mess in this topic within my posts, however, as far this topic run, as easier it is to me to come with my own right conlusion.

The same luckly this time.

I think I`ve solved my problem mentioned in last post. I`ve found even a way to sketched spline within a curvature in any place from previous splines(created earlier)

This is the base for my case



The goal was to make another spline in place between those two references, and maintain it with the same curvature as spline shown above.

Well I`ve been strugling with this a little coming with a feeling that a Pro\e misses some analysis tool - in basic foundation at least.

So taking at start a equation mentioned by jeff in another post

r = 3 * a^2 / b / 2

I needed a radius in intresting me point. It is possible to obtain this data and store it as feature, however only with BMX license, which I miss very much.

I`ve made some workaround taking advantage of evaluate feature. It can provide info about radius in specific point, but it can do this only for curvature. Having this stored as parameter there was only one realtion left to write

r1 =rc:fid_radius_c

where radius_c is measured before value

rest was quite easy to set



sd11=3*sd10^2/rc/2

the curvature was obtained, however the plot was not smooth

To finish this I created special sketcher with a line to measure its lenght by evaluation feature



with this, there was only one relation left to write in final sketch

sd10=l1/2

where l1 is the lenght from earlier pic. I`ve noticed that dividing l1 by 2 gives the best results(still I don`t kno why it should be 2)



I`ve hope it wasn`t boring)))))

there is for sure easier way to solve this case and I looking forward to see it

 

[design-engine]

I will try to dig up that Alias model. I do not teach that class so It could take me a little bit to find the Alias model. Then Ill export it and upload her.

Edited by: design-engine

 

[jeff4136]

> I will try to dig up that Alias model.


Super deal. At your leisure and it will be appreciated.
It's alwaysinformative to see how the 'real guys' do things.


> ... and upload her


If they ever sort out what they messed up yesterday with the
'server upgrades'.

Edited by: jeff4136

 

[design-engine]

I uploadeda Claybake challenge solution from http://www.productdesignfourms.com set ofvideos to youtube.com ISDX and Pro/SURFACE sample of surfacing techniqes specific to design-engine


http://www.productdesignforums.com/index.php?showtopic=8141


Check those out and don't make fun of me. My first few sets of vids.


http://www.youtube.com/bbrejcha < there it is
Edited by: design-engine

 

[jeff4136]

Jacek,
I have some stuff to post but the forum got hosed after the Friday 'upgrades'
and I can't see or post attachments. If they don't fix the forum soon, I can
email the stuff if you'd like.


> Some surfaces behaviour: r1 =rc:fid_radius_c


It's not apparent; are you aware that [curvature * radius = 1]?
So if you'll create your Evaluate feature measuring curvature you
can create your parameters = 1 / curvature = radius.


> G2 surface problem (ISDX):
> map the curvature on a relatively simple surface edge then drive a
> curvature dimensioned spline with an Evalgraph feature


This can get pretty tricky and is always labor intensive as compared to
creating a G2 Boundary Blend between two edges. The results can be
interesting (good), especially if you use a Bezier curve (spline with
four CV's) but creating the control polygon constraints is time consuming
and the range within which a Bezier can be constrained is limited. The
relations for one of the example files ...


/* Given r = 3 * a^2 / b / 2
/* and curve with CV0, CV1, CV2, CV3
/* "a" represents control polygon segments
/* CV0 -> CV1 (for one end of curve [sd17])
/* CV3 -> CV2 (for opp end [sd22])
/* "b" represents the offset distances from
/* those segments [sd19, sd23]
/* Values for "a" must be calculated such that
/* there is convergence a(1) -> b(2) and a(2) -> b(1)
/*
/* Initial setup: get it close ...
/*sd19 = 3 * sd17^2 / kd0 / 2
/*sd23 = 3 * sd22^2 / kd1 / 2
/* ... then comment above and ...
solve
3 * sd17^2 / sd19 / 2 = kd0
3 * sd22^2 / sd23 / 2 = kd1
sd22 = kd2 - sqrt(sd19^2 + (sd19 * tan(90 - kd28))^2)
sd17 = kd3 - sqrt(sd23^2 + (sd23 * tan(90 - kd29))^2)
for sd17 sd19 sd22 sd23


... illustrate the complexity (and one of Pro/E's "neat" but seldom
demonstrated features).


- - - -
P.S.
Correction: "the range within which a Bezier can be constrained is limited."
should read 'a degree 3 Bezier'. A degree 5 Bezier (having six CVs) can be
constrained with an infinite range of curvature_end_1 and curvature_end_2
values.
Edited by: jeff4136

 

[muadib3d]

:r1 =rc:fid_radius_c


It's not apparent; are you aware that [curvature * radius = 1]? So if you'll create your Evaluate feature measuring curvature youcan create your parameters = 1 / curvature = radius.

yeah, you`re right, without a doubt. It should be corrected but I missed it - r1=1/rc:fid_radius_c

I have the same in my relations.

I discovered, that unfortunately, I am not the first one who found that trick. sh*t, I was so happy that I figured out it by myself))

I noticed the same trick in Your model Comb_vs_doodle, but then I didn`t pay such attention to all rels included in sweep profiles or sketchers.

the rest of given in your previous post relations must wait till I find some more time to play with them a little.


Bart

I gonna take a look at Your Youtube movie.

I found it is really hard to get some intresting movies considering surfacing\modeling stuff. There are some for Rhino, more a or less for Alias(I see those for Rhino more worth than for Alias). 3ds or Maya are out of my scope, becouse of lack of curves usage.

I found also some really intresting tips on 3CADtips about filleting in Rhino. Ihaven`t suspect that I could find there anything intresting, and then - such surprise! ))