Hi Group,

I have been trying to build a smooth shape in MoI today but not quite ending up with
what I want......take a look at the attached file.

Around the origin is my own technique for building a smooth form using those curves but
after making surfaces they will not join. I am not sure what the problem is, wether it is
tolerance related or otherwise but im at a loss why MoI will not join them for me. The
'other' solid above was made in a different app but using the same curves......

--

I would like to understand the problem and workaround (if there is one) but I would also like
to see other ppls suggestions on how best to make a smooth surface out of these section
curves ???

Rgds,

Hi Paul,

I am a little confused, you show 3 different models (similar, but with different control).

Anyway, to make a surface on the "please try", I would split the closed base polyline, then select all and network. The resulting surface will only be as good as the control curves.

Hi Paul, they do seem to be just very slightly out of tolerance - there is a distance of 0.009 between them, and MoI will currently join up to 0.005 .

Which surfacing command was used to create those pieces, I mean was it a sweep, network, something like that? The likely bug that needs to be fixed is to add a bit more checking to ensure curves have been fit to the 0.001 tolerance properly, the refitting procedure let through a very slightly sagging fit curve in this case which is basically the root of the problem.

One interesting thing is if you shrink the model down by a factor of 0.1 (select it, run Transform/Scale, type 0 <enter> to pick origin point, then type 0.1<enter> for scale factor), it will then join because then the gap is within the max join distance.

If you can let me know which surfacing command was used for building those pieces, I can see if I can repeat the problem starting from the initial construction, that will give me the best chance to solve the fitting problem.

At some point I will probably be trying to overhaul the way that tolerances are used in MoI in general and try to focus more on adapting the tolerance to the scale of the object instead of having it fixed at an absolute value of 0.001 . I'm not quite sure how that will work yet or when I will tackle that though.

- Michael

Hi Steve:

Thanks for you reply. Sorry to confuse, the top model was built out of MoI but using the curves and structure
built in MoI, I was trying to demonstrate that I didn't end up with a gap and that the curves were OK.

The middle surface with the curves was built on MoI but could not be joined up which was my main reason for posting.

The lower surface was to try and get some feedback from other experienced users of the best way to model a smooth
surface given those three curves.....This style of shape typically breeds the dread 'X' in the surface with bulged corners
as can be seen when you use the technique you mention.

>> The resulting surface will only be as good as the control curves.

So how might I improve this in your opinion.....the only G1 curve is the profile/split ??

--

Hi Michael,

>> Which surfacing command was used to create those pieces, I mean was it a sweep, network, something like that?
>> The likely bug that needs to be fixed is to add a bit more checking to ensure curves have been fit to the 0.001
>> tolerance properly, the refitting procedure let through a very slightly sagging fit curve in this case which is basically
>> the root of the problem

Curve network for both surfs - HTH

>> At some point I will probably be trying to overhaul the way that tolerances are used in MoI in general and try to focus
>> more on adapting the tolerance to the scale of the object instead of having it fixed at an absolute value of 0.001 .
>> I'm not quite sure how that will work yet or when I will tackle that though

I think MoI's tolerance in general is quite tight, we normally model at 0.001mm and comparing offset curves with R, I find
MoI makes quite noisy curves in comparison...

Best Rgds,

Hi Paul,

>>I was trying to demonstrate that I didn't end up with a gap and that the curves were OK.
>>Curve network for both surfs

I have not yet seen gaps left in network within MoI, or I would of posted. Could you please post your construction to replicate that.

>>This style of shape typically breeds the dread 'X' in the surface with bulged corners

Bulged corners/surfaces can be a problem, but it is initial from the input. Yes, some software will attempt to compensate, and make attempted correction, but I do not expect that from MoI

>>So how might I improve this in your opinion.....the only G1 curve is the profile/split ??

Give me more info on the resulting component needed. Simply showing 3 curves and expecting perfect smooth is difficult.

- manz

Hi Steve,

>> Could you please post your construction to replicate that.

Open the file I attached and try to join the two surfaces on the model in the middle (about origin) ??

>> Bulged corners/surfaces can be a problem, but it is initial from the input. Yes, some software will
>> attempt to compensate, and make attempted correction, but I do not expect that from MoI

Well, in fact MoI's method does in fact make sweet surfaces on the whole.

>> Give me more info on the resulting component needed. Simply showing 3 curves and expecting
>> perfect smooth is difficult.

Nope three curves is all you get to define the shape, you do however get a poetic license to do whatever
you want in between to make the surface as smooth as possible ??? now that beats the Jeep challenge!!

Cheers,

Hi Paul,

>>Open the file I attached and try to join the two surfaces on the model in the middle (about origin)

But how did you construct those surfaces?

>>Well, in fact MoI's method does in fact make sweet surfaces on the whole.

There are some problems, but for the expected user, then I think most should be OK.

>>Nope three curves is all you get to define the shape, you do however get a poetic license to do whatever you want in between to make >>the surface as smooth as possible

Interesting. Is there limits/avoidance on that?
MoI does not include checking for that (deviation etc), but I am interested on what you attempt and expect from such software.

I could say,.. use solidworks,... or solidedge,.. I know and use them, but that is work for me. MoI can be a pain sometimes due to some limitation, but come on,... for the price,... I purchased license not based on how it is now, but more on how Michael will continue.


- manz

Hi Paul,

Sorry, bad day yesterday.

I have taken your surface (the one you built in "other"),.... created the surface in MoI based on the curves you posted ("please try" curves), and created the surface in Rhino 4(with those curves).
This is the output for surface analyze in Rhino 4. You will note the surface you created in "other" is in error (there is a kink or crease), as the surfaces match to position only (G0), from MoI and Rhino the surface looks much better for curvature and tangency and looks G2







- manz

Hi Steve,

>> But how did you construct those surfaces?

Curve network for both surfaces - read my reply above to Michael.

>> There are some problems, but for the expected user, then I think most should be OK

I would interested to hear about the ones you come up against the most, hopefully I can
avoid that situation ;-)

>> Interesting. Is there limits/avoidance on that?
>> MoI does not include checking for that (deviation etc), but I am interested on what
>> you attempt and expect from such software

Given the shape in my example there is only so much you can do between the section
curves anyway, so long at the shape looks smooth that is all im looking for. Actually, that
is not entirely accurate as there are situations whereby it must meet a certain volume too
but we won't go into that here....

>> I could say,.. use solidworks,... or solidedge,.. I know and use them, but that is work
>> for me. MoI can be a pain sometimes due to some limitation, but come on,... for the
>> price,... I purchased license not based on how it is now, but more on how Michael
>> will continue.

Okay....your'e getting your own back now ? was that intended or me ?

Many thx in advance,

Hi Steve,

The error you talk about is more to do with continuity right ??, I did actually match & improve on G0 in Rhino but I wasn't
happy with the result due to the surface isocurves going to a single point, thats where I lost the control.

Building the surface as one is usually the best option in terms of overall smoothness and your first comment the other day
was 'spot on' as splitting the profile helped make a good surface as shown in your analysis. I say good because if you turn
the surface and look end on there is a crease when the surface again meets at a point.

Are there any other ways to build such a shape or how might it be possible to further smooth your example ?

Many thanks in advance,

Hi Paul,

>>but I wasn't happy with the result due to the surface isocurves going to a single point

Split the base control curve into 4(curve split at tangent/intersection) then network in MoI:-



This is the surface produced in MoI but viewed in Rhino:-




- manz

Hi Paul,

>>I say good because if you turn the surface and look end on there is a crease when the surface again meets at a point.

What did you use to check the surface? If you are just looking at the model in MoI then you may just be seeing an error in the created mesh not the actual surface.

- manz

This is my little example.
I took the MoI file as an .obj file directly into Carrara.
Most surprised how many of the various choices for networking would not or were not good!

Hi Steve,

Before I did as you suggested in your previous post I took the model into Rhino and shaded it, if you look you
can see a bad crease on the end.

This method of further splitting the curves is an excellent idea !! Thanks

There is one small area which still doesn't look quite right and im not sure how that could be resolved. Looking
at the profile of the surface you can see that the iso curves don't quite make it to the end of line and it is this gap
to the arc where the deviation is taking place.....this would also be a pain to machine in the area.

If you have any further ideas I would like to hear them. :-)

TIA

Hello Brian,

It's hard to say but I don't think your surface follows the curves both directions ?

Hi Paul,

Move the splitting of the base control line to the ends,.. it does less pulling on the end areas.:-




If there is still a problem, then I would suggest a surface "rebuild" in Rhino.

Hi Steve,

Perfect! - I have learned a lot using the network command in MoI and the various splitting of curves to make major
& subtle differences - Thanks for your time on this one.

I will be cutting this model in the morning :-)

karter
Is this ok?
(It's the same file with a different render setting._
Brian

ps. The MoI screen shot shows the only way I could get a nice smooth network result. Doing a left and right version and joining them.
(Sorry if I am missing the point somewhere but could not my end result be as clean/smooth as is required.?)

Hi Brian,

Thanks for the reply. If you look at the previous posts and Manz method of splitting the curves in a particular manner to create a network surface that works the best...try it and see ?

Building half or even quarter surfaces with limited geometry like I have requested will definitely cause problems with continuity after being mirrored into a full shape...maybe V2/V3 will have some matching capabilities that would help this cause.....

Thx again.

You must be cautious to use Network for this type of surface. When input curves are G1, the resulting surface might seem OK and will be valid. BUT there is usually a problem with the surface normal at the corners. Try to Offset the surface and see what happens... If you're not planning any downstream operations that will confilct with this (shelling, fillets, etc) then it probably won't be a big deal...

jonah