modeling a car

[jeff4136]

View attachment 741

2005-03-19_144718_crv_graphs.prt.zip

.... might help clarify "G2", etc.

"Smooth" is a toughy; subjective, a blend of almost art and technical ability (using available tools to define and create). Myself, I'm just stumbling through the fog trying to figure it out. One of the things I think I "know" about the subject; the most common error is trying to "overdefine" curves and surface; add control points to curves and sections to blends, etc. There's a lot that goes into it and very few "one size fits all" answers.

WF2 has an excellent set of analysis tools for evaluating surfaces and curves, as well as an excellent set of definition / creation tools. About the only thing I miss, and rely on Rhino for, is being able to see the actual isoparm structure, assuming the ability isn't under my nose and I'm missing it. Directions can be seen, but not the actual density, as far as I know. One of the most useful evaluation tools (don't overlook it) is the real time rendering and environment mapping. Look at the way light and reflections play across the surfaces and intersections.

Rhino is, IMO, an excellent tool for figuring some of it out, evaluation, etc. Really good program. The more I learn about the tools available in WF, though, the less I rely on it for shape creation; not to say I'm ready to put it away just yet.

 

[kujur11a]

Thanks Jefffor ur guidance. It was very nice to learn about the G1,G2, G3 curves. how can we creat these kind of curves in Pro/E 2001. I have seen WF 2.0. Yes there are some good surface analysis options like reflection & others to check the smoothness. But can do that in Pro/E. I think PTC has to work on it to make Pro/E more strong in surface modeling.


kujur

 

[grendel]

it's the first time i come here ,and i find this is a wonderful place.You guys above are such accommodating folk.

ok,about ISDX,i have a question:when we creat a complex model such as a car by useing style, then it is a single feature, and if it is easy to be modifiedjust like what we do in a solid-creating model?

 

[design-engine]

View attachment 760

the trace sketch options. Note the imported jpgs from the US Patent

website. This is a tutor from our level 7 G2 A-CLASS surface modeling workshop.

http://www.proetools.com/courses/isdx/level7.htm



It is easy to modify, althought it takes proper direction and specific techniqes

and practice.



Edited by: design-engine

 

[jeff4136]

... This part has exactly 1006 features, 36 MB Pro/E file.

Huug, that is a nice looking model.

A few questions, if I may...

Approx how long does a rebuild / regen of the model take? Processor and RAM on the machine?

Is the 36 MB the saved on disk size? Saved with shade mesh or wireframe?

Something I've wondered about: is there any practical difference in Merge vs. Join when working with surfaces having coincident edges (no trimming required)? I've gotten the impression there is (just from observation), but am not sure it isn't just my imagination.

Thanks,
Jeff

 

[keith faskin]

hi


although modelling a car is a highly complex project to take on its real emphasis is in creativity not in engineering, yes you can build a car surface within proe to a higher standard than alias, it will not teach you the skills assosiated with proe or CAD. the main reason for takng on a project has to what do you want to learn. proe is a solid modeller with a surface plugin thats functionality has been greatly reduced, it is limited in its abilties and only if you have an already thourough understanding of proe, entering into the style isdx plugin will only frustrate you. the answer is in its name pro ENGINEER.

 

[Huug]

There's a buggy datum curve in the file that must be resolved every time the EOP marker is pulled down the list. If anyone has any clues as to the problem I'd appreciate hearing 'bout them.

Since this is a flat curve on the XY plane, you could sketch it.


Huug

 

[Huug]

> Huug, that is a nice looking model.

Thanks! Altough the project was a disaster..

> A few questions, if I may...

YOU may indeed!

> Approx how long does a rebuild / regen of the model take?

Long! Approximately 4 minutes. The complete assembly did about 18 minutes when changes to the skeletonpart were made...

Took a lot of coffee those days...

> Processor and RAM on the machine?

That was on my old machine, a Dell Precision 330, P4 266, 512 RAM, on ProE 2001
Now on my new machine, A Dell 370, P4, 3.4, 1 GB, WF2, it's about 2 minutes for this part. What sucks is that WF2 keeps all features selected after a Resume Last.

> Is the 36 MB the saved on disk size?

Yes. With the compressed file option set to yes.

> Saved with shade mesh or wireframe?

That doesn't mather. Although working in wireframe or hidden line was quite impossible. Creating the drawing with several views and sections also. I did become patient, though.

> Thanks, Jeff

Youre welcome!

Huug

 

[jeff4136]

Thank you, Huug. That's an interesting perspective on the performance. That's definitely a good example of a "complex" part.

Re the buggy curve: If I remember correctly I used very few, if any, 2D sketched curves for blend boundaries. That was an exercise, experimenting with the tools (Foundation), trying to get G2 blends throughout, and I don't believe there is a way to define G2 continuity with a sketcher curve (is there?).

Thanks again.

 

[design-engine]

anyone know what A-class surfacing is beyound G2 continuity? I am prepared to post the answer.

 

[james.lynch]

by all means please do! interesting one...

 

[BigJoe]

I don't know Bart. Does it have something to do with manufacturability?



Ok, I cheated and did a
<a href="http://www.google.com/search?num=30&hl=en&lr=&client=firefox-a&rls=org.mozilla%3Aen-US%3Aofficial&q=a-class+surface&btnG=Search" target="_blank" target="_blank">Google
search</a> and
<a href="http://www.eng-tips.com/viewthread.cfm?qid=21173&page=9" target="_blank" target="_blank">found
this eng-tips.com discussion on A-Class surfaces.</a>



A little further down the Google search results;<a href="http://www.design-engine.com/stories/classa.htm" target="_blank" target="_blank"> Pankaj's Class A
surface article on design-engine.com. </a> Very cool.
I'll let Bart give the details in this thread if he wants. Click
the link if you just can't wait.



Edited by: BigJoe

 

[design-engine]

A-Class surfacing is defined as these two major points


1 lightly created geometry and that the geometry is built correctly.


2 light reflections are reflecting in a desirable way. i.e. if g2 is required then the light reflects in a g2 way.




A Class surfaceses are those aesthetic free form surfaces, which are
visible to us (interior/exterior), having an optimal aesthetic shape
and high surface quality.




Mathematically class A surface are those surfaces which are
curvature continuous while providing the simplest mathematical
representation needed for the desired shape/form and does not have any
undesirable waviness.
< this waviness is found with a goussian anaylisys.


Curvature continuity: It is the continuity between the surfaces
sharing the same boundary. Curvature continuity means that at each
point of each surface along the common boundary has the same radius of
curvature.







+Compare G0


positional conections. Not tangen but touching.





+Compare G1


simply tangent. The surfaces in this example are tangent, the curve
plot is not tangent. This line can not be broken down to a more simple
curve therefore it is not curvature continuous or G2. You can not take
an integral at the intersection of the curve as it crosses to the other
surface.





+G2

curvature continuous. When a curve plot is tangent. The equation of
the curve plot can be broken down into a more simple curve by taking an
integral. ie. calculus





five ways to determine g2 continuity.


1. Evaluate with your eyes the light reflections


2. Evaluate the math using Calculus
(if you can take an inergral at a curve that crosses a surface intersection then there might be g2 continuity)

3. Look at the curve plot in either Pro/E or Alias


4. Spans or CV's line up in Alias


5. zebra stripes




Some people will use a Gaussian analysis incorrectly. In this tool
G does refer to that specific mathematician however the standard
Gaussian analysis refers to concave vs convex.
ie. when light reflects back to they eye.






http://www.design-engine.com/stories/classa.htm









Edited by: design-engine

 

[jeff4136]

I thought Gaussian curvature was the product of curvatures in the U and V directions (or some such); e.g. a measure of compound curvature. Does it apply to Class A surface evaluation in other than a very indirect way, i.e. looking for abrupt changes?

 

[james.lynch]

What are your thoughts on dihedral angles?


are the five ways that you listed in any particular order? importance? should they all be carried out?


James
Edited by: james.lynch

 

[design-engine]

are the five ways that you listed in any particular order? importance?
should they all be carried out?

Those are not listed in any particular order however the top one is my
favorite. If a shinny ameba shaped light reflection cans slide across a
surface without itself loosing tangency (at the reflection level) then the true
geek in us gets excited.


Zebra stripes are the most useful for evaluating the
surface connections and the curve plot is the most useful for evaluation of
curves.



Dihedral angle
- the angle formed between two bonds on neighboring atoms in a Newman
projection. Hummm?


The Dihedral is that octogon looking lattace structure that if you
have enough facets it looks round. Who was the toy company that
had a dihedral ball for kids? Software programers have to use
these words
when describing polygon renderings (speculation) -shade the
screen and other conections. If the angle between
each facit is small enough
then... it is tangent to the other.



http://www.pence.ca/~jrainey/dihedralcalc.html



When you measure the dihedral angle you want the result to be less than .1 I
think in order to be considered tangent. You measure this dihedral angle
on the connection between two surfaces.



What do you all think? I did not fire up pro but the min max dihedral angle
number should be zero if it is perfectly tangent. I have not messed with it by
adjusting accuracy but I suspect you could get a lower result (better boundary
condition) if you tightened up on the accuracy.




I prefer to just sand the tangency problems out in the mold.



Edited by: design-engine

 

[design-engine]

I thought Gaussian curvature was the product of
curvatures in the U and V directions (or some such); e.g. a measure of
compound curvature. Does it apply to Class A surface evaluation in
other than a very indirect way, i.e. looking for abrupt changes?

If you use a composite curve ontop of R value curves you will foruce G2
but you will never acheave A-Class ontop of that curve. Your
gaussian plot will show a ripple of colors that will reflect from one
spectrum to another then back again. This indicates that the
relfection will relfect an ripple. I think of gaussian as a
quick check so I don't add bondo to the prototype. (that happend
once). It was a big TV set and they built the protoype and it was
smaller on the screen. I learn from my
mistakes.



I am not sure what compond curvature might be refernce for but I think the above covers it.





R value - Curve plot is equal distant from the curve entire lenght of curve. Same radius.

 

[james.lynch]

Thanks for the explanation.. very helpful.. I really need to look more in depth into the surface analysis side of things.. a project for the future!

 

[jeff4136]

... I am not sure what compond curvature might be refernce for but I think the above covers it. ...

Compound curvature is an indicator of formability but I think you mean "within the context of Class A evaluation", and it does cover it.

Thanks a million for sharing the information.

If I might beg a little more free insight; I've found Rhino's spherical environment map function very handy for evaluation and am having trouble getting the same satisfaction from Pro/E's real time rendering, drm setup. Do you have a favorite setup for this purpose (you mentioned an amoeba shape)?

Edit:

... I have not messed with it by adjusting accuracy but I suspect you could get a lower result (better boundary condition) if you tightened up on the accuracy...

I have observed this. Not very big differences and I'm not sure if it's due to recalc of the derivatives or elimination of "noise" induced by the edge mismatch.

Edited by: jeff4136