We're having a "party in a box".........I think we'll leave the other part out!

BurrMan, you have way too much time on your hands - lol :)

Martin

""""""""""""""BurrMan, you have way too much time on your hands - lol""""""""""

Actually, no. Thats why no modeling and just a party slapped together. (Although I did flow the little guy, in honor of majik.)

Wellll I like it!!! ;-)

There's no denying it, what better way to chillax away a hard day than with the MoI Brand Portable Luau System!

Hi Mike,

> Michael, these surfaces said they were joined, but upon closer examination, <...>

You can have surfaces that are joined together into a larger connected chunk of an object, while still having some various particular edges in there not joined to one another.

Here's an example to clarify:




That object is a "joined srf" because a "joined srf" just means that the object in question is made up of more than one surface joined together.

In this case the joined edges are these ones here:




This area here on the other hand did not get joined because there is too much of a space between the different surface edges there:



So that's the same kind of thing that you have there.

When the object type is "Joined srf" it doesn't mean that every edge is joined - in fact it actually means the opposite of that, that only some edges are joined and some are not, because if every single edge is joined to another one that will be labeled as a "solid" instead of "joined srf".


> when you separate the whole object set and try to join the surfaces
> in question separately, they refuse to.

This usually means that the particular edges have too large of a gap between them - the edges must have a maximum distance of only 0.005 units between them in order for them to get joined together.


> As I have discovered, Loft uses some type of interpolation, which means that the new
> Lofted surface intended to become the sides may have been an in-approximate
> representation of the large surfaces they were created to link.

By default Loft will do an automatic rebuild on each of the profile curves, that can generally help to even out the parameterization of the profiles before they become surface-ized and also to give all the profiles an inherently compatible parameterization which helps to make the final Loft result not be so dense with knots and control points. You can disable that by using the "Profiles" option in Loft - if you use Profiles: Exact then there won't be any interpolation/approximation step at all and the loft will go exactly through the profile without any rebuilding involved. But the Exact mode also means that the resulting surface's knot vector will contain the combination of the knot vector from every single profile curve together and also the surface will inherit any non-uniform parameterization so things like control point spacing can cause the loft surface to bunch or stretch.

But the rebuild step is supposed to rebuild to a tolerance of 0.001 units (unless you have set # of points mode of the profiles which does a much less accurate but also less dense resulting approximation), so it is supposed to be accurate enough to give a joinable result. If that's not happening then that may be a bug in the curve fitter - is it possible for you to boil it down to a simplified example of that problem, like just a file containing 3 curves in it and lofting between 2 of them at a time to generate a surface then does not join? If I had a reproducable example like that then I could probably verify if it is a bug or what.


> I was under impression that Join was good a working around these "near misses".

It is, up to a gap size of 0.005 - if your gap size is somethign like 0.006 units that's still pretty small visually and hard to see the gap without zooming in closely to the area where it is the largest.


> As you can see here, when exported, the polygons are now "matched" together and
> connected with no peering seam.

So that means that there is now a proper joined edge there now.


> Michael, I'd like to know a little about how Boolean Union lined these two edges
> up. Maybe they actually crossed?
> I invite you to also look at those surfaces in the provided model.

I'm not really following exactly what was booleaned. There does not seem to be any model attached to that particular message - in order for me to really follow along with what you are describing I'll need to have a more simplified result, like 2 objects set up to do a boolean union, then I can do the union and see the result rather than trying to reverse engineer some final result model for example.

Right now without simplified examples I just don't quite know the precise answers to your questions because it's hard for me to figure things out just looking at some big pile of surfaces - I don't know where to look or which particular spot the problem is in, etc.. etc...

So if you can simplify things, like "here are 3 curves, when lofted in pairs they don't join", or "here are 2 objects, when booleaned unioned can you give me some more information about the result in this spot here", etc.. would make it so I could better understand stuff and give more information back.

- Michael

Hi Mike, also is it possible that instead of Loft you used Network on the things that don't join and you're also using version 2?

The Network command also does an automatic rebuild of the input curves, but in v2 it did it to a lower tolerance which could sometimes make results that were just a little bit out of join tolerance. That should be tightened up in v3 though.

- Michael

Hi Michael,
"""""""""> I was under impression that Join was good a working around these "near misses".

It is, up to a gap size of 0.005 - if your gap size is somethign like 0.006 units that's still pretty small visually and hard to see the gap without zooming in closely to the area where it is the largest.""""""""""""""

So then when using this method to join "pretty close stuff" would it then succed if you used a scaling workflow and scaled "down, then back up" after the join?

Kindof like the opposite of getting a fillet in a very tiny area by scaling up 10 times.

Hi Burr,

> So then when using this method to join "pretty close stuff" would it then
> succed if you used a scaling workflow and scaled "down, then back up"
> after the join?

Yeah usually that does work, because when the geometry is scaled down the gaps between edges are scaled down as well and that can put them under the join tolerance at that point.

You don't want to scale it down too much though because Join does use the relative tolerance mechanism so things below 1 unit in size in their bounding box diagonal will use a fraction of that diagonal size rather than an actual fixed 0.005 unit value.

So you'd want to scale down by 1/10 in size probably and then scale up after the join.

- Michael

>>MoI Brand Portable Luau System<<

That's a GREAT definition Mike !!This case Burr leave "the Professor" at home and let us discover another part of him.Great idea..just a thing: no coconut trees in your image??...it's a "must" !!

--------

Lejan:that's a PRO image !! Smart use of DOF,box blurred,Yoda on focus...definitely a PRO stuff...

Michael,
Thanks for the detailed explanation! I did not know that about the 0.005 threshold tolerance. That could be used for some good later.

Sorry about that, here is a simplified file: http://k4icy.50webs.com/tutorials/surface_join_area_ex01.3dm
The Orange and Green surfaces are the ones in question.

And honestly now, I can't say that I remember exactly what method or workflow I used to create that set of surfaces.
I may have started with an extrusion of the gap-seam shape that was run the length of the box from end to end. Then I may have performed a Boolean Intersect of the extrusion using a closed curve set created from the inside lip of the box.
Hmmm, and I can think of another way or two that set may have been created.

It's great to know that there are so many ways to skin the proverbial cat.

But with the provided file, you can zoom in close and see that the two surface don't exactly meet.

Hi Mike, thanks for the file - but without being able to also go through the steps of constructing the surfaces, the only information I can really give you is that "yup, you're right they don't meet" - the reason why they don't meet has something to do with the specific method used to construct them. It could be a bug there, or like I mentioned if you use Network in v2 it can tend to make things that don't meet in this type of way but that has been tuned up in v3 already.

According to the curve deviation tool in Rhino those edge are 0.007 units apart, so yup they're just very slightly outside of the join threshold.

If you do the scaling trick as mentioned above to scale both surfaces down by 1/10 in size, then they will join though because then the gap between them is only 0.0007 units and below the join threshold. To do the scale select them both, then run Transform > Scale, at the first prompt for the scale origin type 0 and push enter to specify 0,0,0 as the scale origin point, then at the second prompt type in 0.1 for the scale factor and push enter - now with your objects scaled down they will join and you can scale back up in the same way with a scale factor of 10.0 this time to restore to original size.

I would need to have reproducable steps for the actual construction of the surfaces in order to give you any more detailed information about _why_ they don't join up, right now looking only at the final result it's pretty hard for me to know what went on previously in the construction to end up with that.

- Michael

Well Michael, lets see what I can do....

Here were my steps: (easier to remember when you actually go through them)

I made these two lines and then connected them with the Blend too.


Then I used the Offset tool to make a duplicate.


And connected the ends with a line.
Then I joined the sets together to make a closed curve.


Then I extruded the shape with the ends capped.


Then I Boolean Intersected this extrusion solid with a shape (the interior perimeter) I derived from an inside offset of the box I created.


Then... oh well..... (What in the heck did I get right this time - or got wrong last time???)


Hmm... here is the file I derived these pics from:
http://www.mediafire.com/?r729m1wblhtxm4h


This is how the closed curve was derived from the inter-perimeter of the box:
After I performed a Boolean Difference on the middle section of the box to create the seam,
I selected the perimeter edge curves and Joined them to make a copy.
Now this curve was not planar as it followed the swooping 3D shape of the seam.
I used the handle-bar selection to make the curve "Flat".
I then used that flat closed curve to perform an Offset that was created towards the middle.
At this point it may be a good idea to "Rebuild" this curve.

My only guess is that something went wrong in this process, like an overlapping curve or something not connect.

Sorry, Michael, this is all I can come up with.

But none the less, this entry may be of some use to other modelers.

Hi Mike, thanks for doing the steps and posting the file!

But yeah it looks like these steps must not have been quite 100% exactly the same thing that you did before, because you've got all totally fine and all joined together solids in this result and not the same kind of problem as you had before.

So unfortunately I'm not really any closer to giving you any more information about your previous question about why you ran into that join problem before.

Thank you though for trying to set up a reproducable example - getting reproducable steps so I can see the problem actually unfold through the actions that caused it does greatly help for being able to figure out what's going on and if it's a bug or something in the geometry or whatever...

- Michael

Hi Mike, also previously you were asking about Loft and whether that had something to do with your problem?

But your new steps you show most recently above don't seem to involve Loft at any stage...

- Michael

Maybe I tried it and it didn't work, then did something else...

Yeah, I often get "modeler's regret" when I notice five steps down, that I could have done something better.

I'll often look back at some of my tutorials and realize that I was glad I made them. My inspiration if keen but not so much my memory. ;-)

A vintage version ... :)

I like how you obtained main profile (made two red lines then BLENDED to have a rounded connection )
Leaving other technical discussions,it's logical that rubber seal must be a solid..it's an object,have a tickness,so a solid.
If not happens,you must try to obtain it,searching where are open parts (this is my workflow)

Thanks again Mike

apparently the coconut tree's are growing iron rigid leaves in only 2 distinct shapes now.... We can still party though.....

...now is really a MoI Brand Portable Luau System (i can't imagine a party without coconut trees..).Great!

----

Lejan:the vintage one surely meets Mike's flavour !