Uroborus Bug

[michallach81]

Hi
I would like to report a bug, and discuss about it.
I came across this bug after I've read this thread:
http://forums.autodesk.com/t5/design-validate-document/how-to-get-rid-of-this-error/td-p/6251046
Error described by @Anonymous is pretty basic, on first sight it looks like self-intersecting body... simple.
@NicolasXu advise seems to be ok, adjust form until you remove self-intersection... But what if it's intended?
Some of you should be able to recall design of an iron, that was used in Fusion youtube tutorials.
What if I would like to extend a hadle of that iron, like this:

That what I thougth, when I saw @Anonymous question, that his design was intendend.
My second thought was about, what if self-intersecting body creates closed volume? Maybe Boundary Fill could work?
To avoid auto-conversion, I knew that I need to go to Direct Modeling mode... I've sculpt an iron, then I've used Boundary Fill tool...
And impossible have happened self-intersecting solid appeared!
I was rather expecting to get just bounding volume.
Let's take a look at Uroborus:

I was even able to shell it!!!
Why I didn't report that to support, and made that thread instead?
Because it might be very useful tool, for designing jewellery or organic ornaments.
For now it requires some additional steps but it's possible, and while it may not be any help in @Anonymous case, it's worth trying.
Here's how to make Uroborus, that won't cause troubles:

Michał

[PhilProcarioJr]

@michallach81

In the interest of a healthy discussion, because I am curious. When would you ever intend to have intersecting geometry? I mean it's not manufacturable, it causes 3d prints to fail, you can't machine it, so why would you ever want it? Even in your examples the end result is one solid piece at the area of the intersection. Now one point you do bring up that I would love to support is better tools for joining an area where there is an intersection of geometry, because right now it is less then productive even though joining the two parts is completely possible. But kudos for this discussion.

[michallach81]

Self-intersecting geometry? Never. Iron example is good, how do you think if I would join instead if extend (trough) that handle, would it be the same shape, even with creased edge?

In general you may want to have self-intersecting t-spline just to get ordinary brep.

[PhilProcarioJr]

I do a conversion on the body to create the reference shape, then do a fit in the sculpt tools. My final result has always produced expected results. Granted I wish there was easier ways to do it but you can do it with the current set of tools. Like I said this conversation does interest me and you definitely have skills that I respect. I think in the end what we really want is the same thing, better tools to handle cases like you have outlined in this thread. The current method of doing things like this is all work around so a solid set of tools to handle cases like this would be welcomed from me.

[kb9ydn]

---

@PhilProcarioJr wrote:

@michallach81

In the interest of a healthy discussion, because I am curious. When would you ever intend to have intersecting geometry? I mean it's not manufacturable, it causes 3d prints to fail, you can't machine it, so why would you ever want it? Even in your examples the end result is one solid piece at the area of the intersection. Now one point you do bring up that I would love to support is better tools for joining an area where there is an intersection of geometry, because right now it is less then productive even though joining the two parts is completely possible. But kudos for this discussion.

---

I think at the end of the day (finished design) you probably don't ever want self intersecting geometry.  But it might be useful to allow it so that certain shapes are easier to create.  In regular solid modelling self intersection is done all the time; it just joins with the existing shape.  Is there perhaps some mathematical reason why self intersection is not allowed with t-splines?

This sort of reminds me of the "zero thickness problem" that Solidworks has.  It will not allow you to create geometry that has a zero thickness.

Fusion however has no problem with it.  Now technically this isn't manufacturable either but you could certainly mill (or drill) a shape like this and get something very close.  What happens at the zero thickness area will depend on the material and process used.  But regardless of that, there are still times when it would be handy to be able to do this (in SWX) because you know that later on the zero thickness area will be removed anyway.  Yet for some reason it refuses to allow it (ever!) and you are forced to work around it.  My guess is that there is some mathematical reason for this that is built into the modelling kernel.

C|

[PhilProcarioJr]

@michallach81

Here I joined a complex surface T-Spline and a pipe T-Spline. I was able to keep both shapes exact even after the two were joined into one piece.

Like I said I believe we do need better tools for cases like this though.

[michallach81]

Phil you still missing my point, I'm aware that we can make transition, and keep the shape... but it comes with price...
First, for quality surface you need to find right topology for transition. Second, complex faces layout is less manageable.
But what's most important t-spline workflow is streamlined, and it tooks a lot of work to make changes after all details are added.
I'm sure you will understand what I mean after you'll try to beat me (yes, small challenge among friends).

Below you'll find my screencast and uroborus model in attachments.
First, try to remove self-intersections (jaw & tail, and one in the middle) by adding some t-spline faces (maybe bridge command? whatever).
On second step, using the same body with all transitions in t-splines, make some "big" changes to the main shape (like I did in screencast).
My screencast last 3 minutes, to create any decent transition in crossing places, I would need much, much more time.

Because self-intersecting closed t-spline body also creates closed volume, I thought that we can use boundary fill to get non self-intersecting brep body...
As it turned out, not without another workaround.

All above is just small misunderstandment, we went in to (maybe because of my english?).
What is really important, is a bug. I was able to create self-intersecting solid body!!!
Even more it's possible not only with t-splines, but also with self-intersecting surface and Boundary Fill... more to come, Thicken tool also could do that:

Thank you for your attention, and have a nice day.

Michał

[colin.smith]

Hello all, 

Here is another way to solve this issue:

http://autode.sk/1qsN3kd

Colin

[michallach81]

Nice Colin, but for your workaround to work someone need to be aware of such a problem.
What I'm asking for is a warning in Boundary Fill tool (and others also), and eventually some solution for that bug.

[donsmac]

First, thanks for your post, I had not used the boundary fill before, but I see now its usefulness to convert a self-intersecting T-spline body into a solid.

I made a screencast basically following the same steps you showed in your first post. 

It does raise a couple questions: 

1. you get an error when attemping to Convert the t-spline body to a solid due to self-intersection. The question is, if the boundary fill can do it why not the Convert function? (albeit, should still come with a warning message that self-intersection has occurred). 

2. since the convert doesn't work, we simply use the boundary fill, but wait, now there's that self-interection in solid form. What is the point of that? Why not have it fill in automatically instead of having to take a few more steps to split the body and recombine it?

[michallach81]

1. Because developers, did overlooked? I mean that they allow Boundary Fill to succeed with creating self-intersecting solid.
2. and that's why I've posted that thread, that if developers gonna come with solution let it be a solution for finding that doubled volume and taking that in to account.

[PhilProcarioJr]

@michallach81

Your right it is a bug and they should fix it and give better information on failures. 

I feel it was my misunderstanding of the real point of your thread. Sorry for that. I use T-Splines in a very different way then I think 90% or better of the users here on the boards use them. T-Splines are my go to for fleshing out ideas quickly, then exporting the control mesh for retopoligization. Then I import that back in and get my finished solid. I am going to post improvements to the T-Spline environment later this week and I hope you will give me your opinion because I value it.