Boolean Union - Not sure why it's not doing it?

Next
 From:  ClosedCircuit
6739.1 
I'm trying to understand what I'm doing wrong here.

1. Problem 1: I would like to join the 34 curves (a mixture of lines, arcs and circles) but MoI just laughs at my shortcomings and refuses to do so.

2. Problem 2: Using these 34 curves, I've sweeped a profile and created 34 solids. I would like to Boolean->Union these solids, but MoI just laughs at my shortcomings, etc.

Finally, is there a direction built into curves (e.g. so that sweep knows where to begin and end)? If so, how can make sure that my curves are all pointing in the correct direction?

Thanks!
Attachments:

Image Attachments:
Size: 331.6 KB, Downloaded: 47 times, Dimensions: 1920x1040px
Size: 429.8 KB, Downloaded: 65 times, Dimensions: 1920x1040px
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Michael Gibson
6739.2 In reply to 6739.1 
Hi ClosedCircuit, I'll take a look at your files later tonight, but right now a quick answer to this question:

> Finally, is there a direction built into curves (e.g. so that sweep knows where to begin and end)?
> If so, how can make sure that my curves are all pointing in the correct direction?

Sweep will start at the end closest to where you clicked on. So to make sure your sweep goes in the direction you want, make sure when you select it that you click somewhere nearby the particular end you want.

Most things in MoI work like that where if a curve direction is important it's determined by which end you clicked closest to.

- Michael
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  ClosedCircuit
6739.3 In reply to 6739.2 
Thanks Michael. It's appreciate.

I've made some good progress with MoI (hence why I didn't pester you for a few weeks), but from time to time I come across something really basic which I just can't get my mind around. Must be old age... ;-)
  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Michael Gibson
6739.4 In reply to 6739.1 
Hi ClosedCircuit,


> 1. Problem 1: I would like to join the 34 curves (a mixture of lines, arcs and circles) but
> MoI just laughs at my shortcomings and refuses to do so.

In MoI you can only join curves with just 2 segments that share a common endpoint, not ones that branching structures like you've got there.

A joined curve in MoI is basically for making longer and/or closed curves, a joined curve is made up of segments but the segments touch each other end-to-end forming just a single directional path with the end result.

That's basically because a lot of curve based construction mechanisms in MoI treat a curve as something that has a just one start and one end point and that follows just one path...

With something like you show there, if you want to process all those curves for something you can just select them all but you would not want to use Join on them because they are not set up with just 2 segments connecting at each juncture area....


> 2. Problem 2: Using these 34 curves, I've sweeped a profile and created 34 solids. I would like to
> Boolean->Union these solids, but MoI just laughs at my shortcomings, etc.

There are a variety of things contributing to that - one being that it's just more difficult in general for booleans to process things that have a whole lot of overlapping surface areas. But you've also got a lot of other things going on here that add further complexity - some of the pieces are not solids like with missing end caps which will make it difficult for the booleans to figure out which pieces are supposed to be kept and also some pieces are made up of joined parts but with self-intersecting pieces which really throw a monkey wrench into the boolean process as well.

Like for instance this piece here:



That's an open surface and not a closed solid because it's missing the front cap piece here:



With it not being a solid closed volume it will behave pretty differently than what you might expect when doing booleans with it, the booleans are more oriented around figuring out which pieces to keep or discard after slicing things with each other based on which solid volume they are contained inside of.

Then in addition to that the self-intersection is here - these 2 surfaces skim over some overlapping surface area:






That kind of self-overlapping surface skins will also mess up things that want to analyze solid volumes.


So there's still probably a fair amount of tune-ups needed on each piece before they'd be possible to boolean union, it may be better to just work using Trim instead which will let you work at a surface level instead of needing solids.

Hope this helps give you some idea of what's going on!

- Michael

  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  Michael Gibson
6739.5 In reply to 6739.4 
So to clean up that part with the overlapping pieces, you'll need to use some object repair techniques, there is a tutorial on this type of stuff here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=446.17

For this particular case you'd do something like this - select it and run Edit > Separate to break it into individual surfaces.

Then select these 2 surfaces (looking at the underside here), and run Edit > Trim:



At the trim prompt to select cutting objects, push "Done" or right-click in a viewport to signal you want to do a "mutual trim" where each object is both cut and used as a cutter.

Now Trim will dice up the surfaces and you'll be able to pick which pieces to remove. But in this case it's easier to pick the pieces to keep instead, so switch it to "Keep" mode and select the 2 visible pieces, there will now be an edge where they were intersected with each other:



Now there's also some overlapping pieces on the top and bottom to fix up as well. Since those are all planar, instead of trimming them with each other instead just delete all the top faces and draw in a new connector line here:



Then select the entire outline and run Construct > Planar to build just one single big trimmed plane surface for the entire bottom, it's better to have things like this as one big plane rather than a bunch of little coplanar fragments:



Then repeat that on the other side and also build a planar cap for the middle side area, then join up all those pieces and you'll have a finished well formed solid for this particular piece after that.

One shortcut is if you build the cap for the middle side area first and then join all that together so that there are planar openings on the top and bottom with the side walls all complete, you can then select the entire joined piece and use Construct > Planar to build planar end caps on it. The Construct > Planar command works either on a set of planar curves, or also if you select a joined surface object it will see if the open edges form a closed planar outline and build a plane and join it in there, that can save some steps.

Hope this helps!

- Michael

  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
Next
 From:  ClosedCircuit
6739.6 In reply to 6739.5 
Hi Michael,

Thanks a lot for your explanations. I understand the next steps which is great. I know what I'll be doing tonight ;-)

At a more philosophical level, is my strategy correct (i.e. trace the labyrinth with curves, accepting that the branching off will trigger some problems at a later stage, sweep and clean up) or should I have approached the whole thing differently?

My thinking is that the labyrinth will be Boolean-Diff'd into a box which will itself be deformed into a cylinder for 3D printing an object similar to the one in the picture below.

Cheers

Phil


  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged

Previous
 From:  Michael Gibson
6739.7 In reply to 6739.6 
Hi Phil,

> At a more philosophical level, is my strategy correct (i.e. trace the labyrinth with curves,
> accepting that the branching off will trigger some problems at a later stage, sweep and
> clean up) or should I have approached the whole thing differently?

It's not particularly a problem to have the initial branching structure, it just means that you can't join the segments together at that particular stage.

Definitely the overall strategy of getting curves to start with is a good direction. The main other way you could do it though would be to stay at the curve construction level for a while longer and build the full thickened outline as curves first before starting to build surfaces. That would mainly involve using Construct > Offset to build curves coming away from the centerline and then probably using Edit > Trim to cut pieces at juncture areas.

So then you'd have an outer curve structure like this:




And an inner one like this:




These outlines now can be joined together to make a larger multi-segment curve. Then you can put them on different z levels and then build a surface between them by doing Construct > Loft. If they have the same segment structure (same number of segments in each multi-segment curve), they can be lofted all in one single go, if they have different structures between them you'd need to do it in pieces at a time like select these 2 pieces and do a loft:






And actually you can save some time by doing it slightly differently than that even - just build the outer sized outline and then use Construct > Extrude with the "Tapered" option, as long as the distance and draft angle you use will not cause different areas to collide into each other.


Anyway, some method like that where you're working a little bit longer at the curve level and building the sort of full outline first before generating a surface would probably be more efficient than constructing a lot of sweep pieces from the centerline since then you have to deal with combining the sweep chunks together.


Hope this helps!

- Michael

  Reply Reply More Options
Post Options
Reply as PM Reply as PM
Print Print
Mark as unread Mark as unread
Relationship Relationship
IP Logged
 

Reply to All Reply to All