Booleans with open surfaces?...
All  1-12  13-15

Previous
Next
 From:  Michael Gibson
3197.13 In reply to 3197.11 
Hi Will also one other note:

quote:
Now I did perform a test yesterday, exporting two joined surfaces into Cinema 4D - all the normals were in the proper direction, but the app I really need to test is Lightwave since that's the one that exhibited the flips the last time I ran across the issue.


Also when you join pieces together, what it will do is give you a _consistent_ normal between each piece.

However, unless it has joined all together into a solid, it is possible that the whole piece may not have the normal direction that you want.

That's something that happen in any 3D app where you are not working with closed objects, including polygon modelers as well - it's why there is a flip function in all polygon modeling programs.

If you have everything closed up into a solid, then again one of those convenience features kicks in and MoI will know how to automatically orient things to the "outside" of the solid. But if you do not have a solid MoI (and any other 3D program really, including NURBS and polygons as well) does not have any good way to know which side is supposed to be the outside.

So if you work with open objects in MoI instead of solids, you may need to flip your object at some point, either in MoI (set up a keyboard shortcut and put in Flip as the command name), or more likely in your polygon modeling program where the normals have become important to you.

But again that is not something specific to MoI - any 3D modeling program that works with open objects may require flipping at some point because it is not generally easy for software algorithms to recognize which side of an open mesh is supposed to be the "outside" part, since there is only a clearly mathematically defined inside and outside with closed objects. Without it being clearly defined, it tends to require human judgment to say which way they are supposed to go and it is difficult to emulate human judgment in software.

But Joining will still help you significantly because it will make the whole joined piece at least have a consistent normal throughout it, so that individual sub-pieces of the joined object do not need to be flipped independently, just the whole thing may need flipping still.

The only way you can avoid the need to potentially do any flipping at all though is to make solids before exporting the mesh - that's the case where it is handled for you automatically.

- 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
3197.14 In reply to 3197.11 
Hi Will, so just an example with flipping...

Say you are in a polygon modeling program and you create one polygon like this:



Then you add a second polygon here:



And then a 3rd one here:




At this point you have a mesh with 3 polygons. It does not necessarily have its normals oriented towards what now appears as the "outside", you may need to flip it. It depends on how the first polygon was drawn and what direction it happened to start out with.

The polygon modeler doesn't really understand that the first created polygon was supposed to be the front face or the back face or whatever of the final object.

The same thing will happen when working with NURBS surfaces as well - it's not something that is a bug to be fixed, it's something that is part of working with open geometry in any kind of geometry system.

So it is not unexpected or abnormal or a bug that you may need to flip open pieces, it's an expected part of the workflow.

But if you work with solids, it is another thing that you can avoid worrying about because it then does get handled for you.

- 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
 From:  jbshorty
3197.15 In reply to 3197.2 
Paolo wrote: "I used SolidthinkingLt and the update(Forma) a few years ago, and it crashed all the time..."

Same experience here. I once spent a few weeks thoroughly testing ST as i wanted a parametric modeler to complement Rhino. To me, it was then (and probably still is) the most similar parametric system to Rhino (and now to MoI) in the overall feeling. But stability was always an issue. Even simple history modifications would often crash it. I will give ST developers some credit as it has some really cool features such as point editing a surface and not breaking history from the input curves. But then there are other things (those things which may be done 20 times in a session) which are so much more complex to do than in MoI or Rhino.

jonah
  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

 

 
 
Show messages: All  1-12  13-15