[Peter Gillespie]

Hi All
Here is a video of a tech question asked here in the Uk About making an irregular shape in Bobcad V27

https://www.youtube.com/watch?featur...&v=sT6Wxdz0DGo

Hope this helps anybody in creating surface's in Bobcad

[magicniner]

Does changing the points chosen for the connecting lines change the shape of the created surface?
Thanks,
Nick

[Peter Gillespie]

Hi Nick
as you can see what I was doing is creating intersection points to create an anchor point to generate the surface
the shape will change but only to the anchor point you create you want to create a surface that's the same at top profile to the bottom profile
the customer was happy as he created what he wanted to manufacture
remember these video's are only a guide to help the concept in Bobcad and give the user a better understanding in Bobcad

[magicniner]

Thanks Peter,
I asked about choice of points because I thought you might have some rules or guidelines to avoid having straight lines cutting across what would ideally be curved?

[The Engine Guy]

Thanks Peter,
I asked about choice of points because I thought you might have some rules or guidelines to avoid having straight lines cutting across what would ideally be curved?

Hi Nick

In the example it has to be done with straight lines as there is no way of knowing what the rads would be of any curves, easy enough to use curves if you know what they should be

[magicniner]

Hi Nick

In the example it has to be done with straight lines as there is no way of knowing what the rads would be of any curves, easy enough to use curves if you know what they should be

I've obviously not made my query quite clear enough, I'm not suggesting knowing what the curves might be or calculating them, I wondered if there was a way to reliably avoid cutting a straight line across what should be a curve.
There will be points that can be chosen to join top & bottom where there will definitely be a straight line, I've not studied enough 3D geometry to know but I'd assumed there would be some way of determining these or at least some guidelines that get you somewhere close,
Regards,
Nick

[Peter Gillespie]

Hi Nick
you can use curves if you know the points
In Delcam they use what is called a pcurve which you can manipulate
Same for Bobcad you can use different points to manipulate the line or curve
its down to the designer to achieve what he wants
Thank you Engine Guy :cheers:

[magicniner]

I will again endeavour to explain again -

The most direct transition between two vertically spaced shapes can be resolved as an infinite series of straight lines, the built in algorithms in BCC do this quite well but mysteriously deviate randomly from the initial shapes and thus the requirement to define lines to divide up the area around which a surface is required.

The reason for my question is that when choosing arbitrary points if you choose the wrong points your straight line will cut across what should be a curve, purely because you haven't chosen the right points.

Having played with some models I suspect that the answer may be that the shortest possible line between the top and bottom shape may be what is required, I've realised that this is a theoretical 3D geometry question which may step outside the experience/capability of members to answer, I just thought someone might know?

- Nick

[BurrMan]

Hi nick,
It sounds like you have mixed 2 questions into 1.

So you are describing how a nurbs surface will be shaped when created between 2 disparate curves. Especially when morphing shapes. The easy answer about where the straight is calculated on peters example is projecting them both to the same plane then creating points where the 2 shapes intersect. Because the skinning and lofting functions will be working in plane.

This would be harder if your shapes had no intersecting points. You would use a curve graph in something like rhino, and you could determine it there, but what you are also asking seems to be a reverse engineering question when you say "avoid cutting straight on what should be curved?" Trying to create a morphed nurbs loft that "matches" something would be hard at best. Here you would look to a program that can do a "fitting process" to some type of existing data, like scan data.

Did i miss your point?

[magicniner]

Hi nick,
It sounds like you have mixed 2 questions into 1.

That's be my inexperience in 3D geometry showing ;-)

The easy answer about where the straight is calculated on peters example is projecting them both to the same plane then creating points where the 2 shapes intersect.

That makes a lot of sense, I was hoping that the point choices in the video would explained as that would help others facing different shapes if they knew the reasoning involved.

I can see the reason for choosing the end points of a curve in the bottom shape but then the corresponding choices on the circle seem to be based on arbitrary geometry - the angle of the lines constructed will change the position of the top points on the circle, so you could simply choose arbitrary points without using construction lines unless there was some unstated reasoning used in choosing the direction of the lines.

Did i miss your point?

Mostly not, and you've given me some useful and substantial points think about.
Regards & Thanks,
Nick

[The Engine Guy]

Hi nick

Yep it all changes when you go from 2D (2.5D?) to 3D. even what we term as geometry in 2D then becomes a "wireframe" and the geometry refers to the surfaces, takes a bit of getting used to.

Anyway, I have to do a lot of surfacing and one way I have tried that seems to be reasonably successful would be in the case of the example used set the points on the upper lines to the same percentage along their length and join at those points, a fair bit of work needed to work out the percentage points along the lower lines but it does seem to be pretty accurate shape wise.

Just another method to try Doesn`t mean it is any good to you

Regards
Rob
:rainfro: :rainfro: :rainfro:

[magicniner]

Just another method to try Doesn`t mean it is any good to you

Au contraire Rob, the more strategies suggested the better and the more reasoning behind them the better,
Regards & Thanks,
Nick

[BurrMan]

I can see the reason for choosing the end points of a curve in the bottom shape but then the corresponding choices on the circle seem to be based on arbitrary geometry - the angle of the lines constructed will change the position of the top points on the circle, so you could simply choose arbitrary points without using construction lines unless there was some unstated reasoning used in choosing the direction of the lines.

Well, the choice of the connections should not just be "arbitrary".... In Peters case, you want them to be "in plane".... But also "normal".... to build quality geometry.... It would really only come into play if you will be doing more advanced operations on it after....

So where you see him creating lines kindof arbitrarily to get connection points with the top and bottom, will work, but really you want to do this with a purpose. BobCads Cad tools don't have really have the command needed, which is a normal or a perp snap, but you can get the result with the arc 3points command... You would select the segment on the bottom at it's endpoint, then select the top arc (the one you are looking to get a perp break on), then select the entity directly connected to the first pick point... Then intersection point where the 2 arcs meet to get the break/attachment point....

The reasoning for wanting to do this in these type of constructions, is to control the underlying topology of the surfaces you create... The underlying UV structure of your surfaces can be viewed by choosing view-wireframe (This will give a simple view of the isocurves of the surfaces....) A well constructed model will have these flow well with each other, from surface to surface as well as in direction with your intention of the surface.... Building those connecter points at arbitrary angles like that will kindof "stretch/stress" the surface where it doesn't need to be... Not necessarily bad. Just not good practice or habit. (of course, unless you wanted that topology to "lean/skew")