The 3dm file Danny originally posted has many duplicate items in it. try to delete all the extras first to have only one of each.

Thanks Anis for the zebra, yes I can imagine continuity trouble with the network surface.
Here's an other way



But I get some trouble with the g1-g2 blend, getting really strange result , I will post it a little bit later.

Hi Danny,

> and what other tools from other (NURBS) apps out there would do this easily?

Really the easiest way to get a very evenly smooth branching structure tends to be using subdivision surfaces. Tsplines is probably the best for that since it will allow you to generate NURBS output.

An example here:
http://www.tsplines.com/community/casestudies/jewelrydesign.html

However, by using that method you also will sacrifice a bit of accuracy because the straight "tube" parts will be part of the melted-down type surface and will not be an exact perfect cylinder shape like you've got in your actual model here.

But if you are focused more on the blended area than the straight parts, then that is probably the easiest way to get very natural "all directional" blending.



Other than that with traditional NURBS tools you probably need to use an "n-sided patch" type tool that will place a fitted surface that is tangent to the input edges (MoI does not have this tool yet).

But it is best to narrow down the area that will be patched to a small region instead of a large region. I would do that using the method that PaQ showed above by doing some blends between the pieces where possible to reduce the area to be patched, like this:



It tends to be harder for that kind of "N-sided patch" tool to do its job if the edges are bending around a lot, often times that tool works by starting with a best-fit plane and then deforming it along the plane normal to fit the shape. The more than you can reduce the area to be fit to be semi-roughly planar will help with that.

Currently in MoI I think that PaQ's method is the best. And actually the zebra results show a pretty good tangent result there.

- Michael

Hi Anis,

> Your tips not tangen but looks good enough....
> Below is the zebra strips :


Actually it looks pretty much tangent to me from those zebra stripes that you showed there.

You may be confusing tangent continuous with curvature continuous.

When the zebra stripes are aligned with one another but have a kink where they meet, that is tangency. You can see the same thing happen on rolling-ball fillets as well. It is when the zebra strips do not even line up with one another that you don't have tangency.

If you want to have stripes not only aligned but also at a smooth transition - that is curvature continuous which is another level of smoothness beyond tangency.

The main thing with that method is that the filled-in patch is actually a planar surface with no curvature to it, so that part is only going to be tangent to the surrounding blends, not curvature continuous.


Can you please post the model of the result you got in SolidWorks? Since it is one surface it is certainly curvature continuous on the interior of itself, but often those tools only produce tangency where they meet with the adjacent surfaces.

- Michael

Hi Pilou,

> Just a question : why the object of the first post can't
> be "explode"? (separate)

You can explode/separate objects that are made up of multiple pieces.

The surfaces in that original post are all just single-surface objects, they don't have multiple surfaces in them to explode out.

Same thing with the curves - those curves are made up of a single segment. Separate can be used to explode a multi-segment curve into individual segments, but when used on a single segment curve it will not have any effect.

I noticed that the original model does have several duplicate pieces stacked up on top of each other, 2 copies duplicated in one spot and 3 copies in some other spots. Is that possibly causing some confusion?

- Michael

Hi Michael, PaQ

>Currently in MoI I think that PaQ's method is the best.
>And actually the zebra results show a pretty good tangent result there.

Yes I agree, that was the final method I came up with as well, PaQ you can use 'Planar' surface also to patch that last bit it is actually flat.

>Other than that with traditional NURBS tools you probably need to use an
>"n-sided patch" type tool that will place a fitted surface that is tangent
>to the input edges (MoI does not have this tool yet).

I Like the way you ended that sentence with 'yet' what other 'yet' tools are you thinking Michael ? :)

Hi Burr,
>The 3dm file Danny originally posted has many duplicate items in it.

Doh! thanks for the pickup Burr, It looks like I left the circular array item count at the default of 8......Corrected!

Thanks to all
Cheers

> Any suggestions Michael?

The way PaQ posted here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=2140.8

is what I would suggest...

It does end up with a planar piece in the top part which does not seem to be an unreasonable result.

If you don't want to have a flat piece at the top there, then it will be a lot more difficult, you'll probably need to use a different program that has an N-sided patch tool in it for that case.

If you really want to emphasize very evenly smooth junctures above all else, then possibly t-splines would be best.

- Michael

>I noticed that the original model does have several duplicate pieces stacked up on top of each other, 2 copies duplicated in one spot and 3 copies in some other spots. Is that possibly causing some confusion?

I have seen that and kill some of them ;)
But that was "the one surface" of half cylinder who pertub me :) (false reflex of the polys of sktechup :)
As Copy / Move lines selected is permitted no problem to use the lines :)

Hi Danny,

> I Like the way you ended that sentence with 'yet' what
> other 'yet' tools are you thinking Michael ? :)

:) A lot of stuff!

Eventually I would think I should get the full set of kind of "traditional" NURBS tools in here...

Some tools tend to be rather finicky and take a while to get working well, N-sided patch happens to be one like that.

But current priorities are a lot more oriented towards the "simple stuff" still. Like I currently have a higher priority on making a better and deeper history function for making basic edits work better, rather than putting a priority on "fancy" / advanced surfacing tools.

- Michael

That's ok Michael as long as those things are in the pipeline, and we know how long the Transcontinental Pipeline is ;) but we can wait as MoI is already a productive piece.

Cheers

Yeah, Paq's method works like'a charm. I even did a shell and it's perfect.

Hi Anis,

>I dont know how to create those shape in moi

Thanks for the demo, is it possible to post an iges, step or .3dm of the Solidworks result so we could have a closer look ?

Thanks

HI Michael, Danny...

Here is what you want.... :)

Hi Anis,

If you examine your SolidWorks result with Zebra stripes across the whole model (instead of only on that one single patch itself), you can see the same kind of zebra line breaks in it, just in a different location where the patch touches the straight parts.





Again though - that does not mean that it is totally discontinuous there - because the zebra stripes line up it means there is tangent continuity there.

- Michael

If you examine that same area with the MoI generated model using PaQ's method, it looks like this:





That same area is very nice with this method.

But instead you get the tangent-only area in a different location - where the top plane touches the blends.


I would think the best compromise shape would probably be to do those side blends but only about 3/4 of the way towards the top instead of all the way to the top. Then fill in the remaining area with an N-sided patch or "fill" instead of trying to fill such a larger area with it.

- Michael

Hi Michael....

Yeah, thanks for checked it....
BTW after object organization done, have you a plan to add some surface or modeling tool for V2 ?

-anis

Hi Michael...

Correct, PaQ method looks better on those area.
So, it will very useful to have fill surface inside moi in the feature.
Network have limitation on making tangency....

Hi Anis,

> BTW after object organization done, have you a plan to add
> some surface or modeling tool for V2 ?

Currently I plan to do a bunch of work on the sweep tools to make those better.

I don't think it is too likely that I will be able to get any brand new surfacing tools in for v2 though. Maybe variable radius fillet, but it depends on how difficult it will be to get a nice UI for that.

- Michael

I did the middle network bit first and then did simple sweeps.

Would that be ok as well?

Brian

Hi Brian,

> I did the middle network bit first and then did simple sweeps.
>
> Would that be ok as well?

It will give a very similar result...

But if it is a goal to ensure smoothness to adjacent surfaces, then Blend can work better for that since it does additional work to create a smooth result the edges that it is given.

- Michael