is there a way to highlight non-manifoldness?

classic Classic list List threaded Threaded
12 messages Options
Reply | Threaded
Open this post in threaded view
|

is there a way to highlight non-manifoldness?

David Goodenough
I have a design which apparently works when you press F5, and also when you
press F6, but when you come to export it as STL, then it fails.

I have reduced it a bit, and I have two major components each of which on
their own will F5, F6, and export as STL, but when you take their difference
it fails.

It would be really useful to have some visual clue to show where the problem
might be.  The nearest comes when I do an F5 on the difference, and an odd
green face appears (for no good reason that I can see) but only from some
viewpoints - move the viewpoint and it vanishes.

Apart from brute force, does anyone have any good debuging techniques in the
absense of visual clues?

David

Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

clothbot
On 2012-05-11, at 9:54 AM, David Goodenough wrote:
I have a design which apparently works when you press F5, and also when you
press F6, but when you come to export it as STL, then it fails.

I have reduced it a bit, and I have two major components each of which on
their own will F5, F6, and export as STL, but when you take their difference
it fails.

It would be really useful to have some visual clue to show where the problem
might be.  The nearest comes when I do an F5 on the difference, and an odd
green face appears (for no good reason that I can see) but only from some
viewpoints - move the viewpoint and it vanishes.

Apart from brute force, does anyone have any good debuging techniques in the
absense of visual clues?

Which OpenSCAD versions have you tried and on which platform(s)?

Andrew.

--

"The future is already here.  It's just not very evenly distributed" -- William Gibson

Me: http://clothbot.com/wiki/



Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

nophead
In reply to this post by David Goodenough
Most often non-manifoldness is caused by CSG operations on coincident faces. There needs to be some slight overlap when doing difference and subtraction and also often with union as well.

I use netfabb studio to highlight the errors if I am stuck and can generally work out the problem for there. I don't get many but I do have a small bodge factor all over the place. Look for eta in the source for Mendel90 for example. It is defined as 0.01 and gets added or subtracted to make sure faces are not exactly coincident.

On 11 May 2012 14:54, David Goodenough <[hidden email]> wrote:
>
> I have a design which apparently works when you press F5, and also when you
> press F6, but when you come to export it as STL, then it fails.
>
> I have reduced it a bit, and I have two major components each of which on
> their own will F5, F6, and export as STL, but when you take their difference
> it fails.
>
> It would be really useful to have some visual clue to show where the problem
> might be.  The nearest comes when I do an F5 on the difference, and an odd
> green face appears (for no good reason that I can see) but only from some
> viewpoints - move the viewpoint and it vanishes.
>
> Apart from brute force, does anyone have any good debuging techniques in the
> absense of visual clues?
>
> David
> _______________________________________________
> OpenSCAD mailing list
> [hidden email]
> http://rocklinux.net/mailman/listinfo/openscad
Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

David Goodenough
In reply to this post by clothbot
On Friday 11 May 2012, Andrew Plumb wrote:

> On 2012-05-11, at 9:54 AM, David Goodenough wrote:
> > I have a design which apparently works when you press F5, and also when
> > you press F6, but when you come to export it as STL, then it fails.
> >
> > I have reduced it a bit, and I have two major components each of which on
> > their own will F5, F6, and export as STL, but when you take their
> > difference it fails.
> >
> > It would be really useful to have some visual clue to show where the
> > problem might be.  The nearest comes when I do an F5 on the difference,
> > and an odd green face appears (for no good reason that I can see) but
> > only from some viewpoints - move the viewpoint and it vanishes.
> >
> > Apart from brute force, does anyone have any good debuging techniques in
> > the absense of visual clues?
>
> Which OpenSCAD versions have you tried and on which platform(s)?
>
> Andrew.
>
> --
>
> "The future is already here.  It's just not very evenly distributed" --
> William Gibson
>
> Me: http://clothbot.com/wiki/
I am running the latest git version on Linux.

David

Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

David Goodenough
In reply to this post by nophead
On Friday 11 May 2012, nop head wrote:
> Most often non-manifoldness is caused by CSG operations on coincident
> faces. There needs to be some slight overlap when doing difference and
> subtraction and also often with union as well.
I have done that, the overlap is 1mm.
>
> I use netfabb studio to highlight the errors if I am stuck and can
> generally work out the problem for there. I don't get many but I do have a
> small bodge factor all over the place. Look for eta in the source for
> Mendel90 for example. It is defined as 0.01 and gets added or subtracted to
> make sure faces are not exactly coincident.
Can netfabb read scad files?  As far as I can see it only reads STL and this
happens when I try to export the STL.

David

>
> On 11 May 2012 14:54, David Goodenough <[hidden email]>
>
> wrote:
> > I have a design which apparently works when you press F5, and also when
>
> you
>
> > press F6, but when you come to export it as STL, then it fails.
> >
> > I have reduced it a bit, and I have two major components each of which on
> > their own will F5, F6, and export as STL, but when you take their
>
> difference
>
> > it fails.
> >
> > It would be really useful to have some visual clue to show where the
>
> problem
>
> > might be.  The nearest comes when I do an F5 on the difference, and an
> > odd green face appears (for no good reason that I can see) but only from
> > some viewpoints - move the viewpoint and it vanishes.
> >
> > Apart from brute force, does anyone have any good debuging techniques in
>
> the
>
> > absense of visual clues?
> >
> > David
> > _______________________________________________
> > OpenSCAD mailing list
> > [hidden email]
> > http://rocklinux.net/mailman/listinfo/openscad


Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

nophead


On 11 May 2012 16:44, David Goodenough <[hidden email]> wrote:
On Friday 11 May 2012, nop head wrote:
> Most often non-manifoldness is caused by CSG operations on coincident
> faces. There needs to be some slight overlap when doing difference and
> subtraction and also often with union as well.
I have done that, the overlap is 1mm.
>
> I use netfabb studio to highlight the errors if I am stuck and can
> generally work out the problem for there. I don't get many but I do have a
> small bodge factor all over the place. Look for eta in the source for
> Mendel90 for example. It is defined as 0.01 and gets added or subtracted to
> make sure faces are not exactly coincident.
Can netfabb read scad files?  As far as I can see it only reads STL and this
happens when I try to export the STL.

No you have to export it to STL. 
 

David
>
> On 11 May 2012 14:54, David Goodenough <[hidden email]>
>
> wrote:
> > I have a design which apparently works when you press F5, and also when
>
> you
>
> > press F6, but when you come to export it as STL, then it fails.
> >
> > I have reduced it a bit, and I have two major components each of which on
> > their own will F5, F6, and export as STL, but when you take their
>
> difference
>
> > it fails.
> >
> > It would be really useful to have some visual clue to show where the
>
> problem
>
> > might be.  The nearest comes when I do an F5 on the difference, and an
> > odd green face appears (for no good reason that I can see) but only from
> > some viewpoints - move the viewpoint and it vanishes.
> >
> > Apart from brute force, does anyone have any good debuging techniques in
>
> the
>
> > absense of visual clues?
> >
> > David
> > _______________________________________________
> > OpenSCAD mailing list
> > [hidden email]
> > http://rocklinux.net/mailman/listinfo/openscad

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://rocklinux.net/mailman/listinfo/openscad

Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

Brad Pitcher
In reply to this post by David Goodenough
I had a similar problem recently. I could export the STLs successfully but they had self-intersecting facets that prevented them from slicing properly. I ended up just commenting out different parts of the design and trying again over and over until I had narrowed down the problem to a particular line which caused to solids to share a face. Once I put a little offset in there it fixed the problem.


On Fri, May 11, 2012 at 8:44 AM, David Goodenough <[hidden email]> wrote:
On Friday 11 May 2012, nop head wrote:
> Most often non-manifoldness is caused by CSG operations on coincident
> faces. There needs to be some slight overlap when doing difference and
> subtraction and also often with union as well.
I have done that, the overlap is 1mm.
>
> I use netfabb studio to highlight the errors if I am stuck and can
> generally work out the problem for there. I don't get many but I do have a
> small bodge factor all over the place. Look for eta in the source for
> Mendel90 for example. It is defined as 0.01 and gets added or subtracted to
> make sure faces are not exactly coincident.
Can netfabb read scad files?  As far as I can see it only reads STL and this
happens when I try to export the STL.

David
>
> On 11 May 2012 14:54, David Goodenough <[hidden email]>
>
> wrote:
> > I have a design which apparently works when you press F5, and also when
>
> you
>
> > press F6, but when you come to export it as STL, then it fails.
> >
> > I have reduced it a bit, and I have two major components each of which on
> > their own will F5, F6, and export as STL, but when you take their
>
> difference
>
> > it fails.
> >
> > It would be really useful to have some visual clue to show where the
>
> problem
>
> > might be.  The nearest comes when I do an F5 on the difference, and an
> > odd green face appears (for no good reason that I can see) but only from
> > some viewpoints - move the viewpoint and it vanishes.
> >
> > Apart from brute force, does anyone have any good debuging techniques in
>
> the
>
> > absense of visual clues?
> >
> > David
> > _______________________________________________
> > OpenSCAD mailing list
> > [hidden email]
> > http://rocklinux.net/mailman/listinfo/openscad

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://rocklinux.net/mailman/listinfo/openscad

Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

David Goodenough
On Friday 11 May 2012, Brad Pitcher wrote:
> I had a similar problem recently. I could export the STLs successfully but
> they had self-intersecting facets that prevented them from slicing
> properly. I ended up just commenting out different parts of the design and
> trying again over and over until I had narrowed down the problem to a
> particular line which caused to solids to share a face. Once I put a little
> offset in there it fixed the problem.
Now it will not let me export the STL, so at least I get an early warning
but tracking down the problem is not easy.  I was just looking for a visual
clue as to where to start looking (a bright flashing red bit would be nice).

David

>
>
> On Fri, May 11, 2012 at 8:44 AM, David Goodenough <
>
> [hidden email]> wrote:
> > On Friday 11 May 2012, nop head wrote:
> > > Most often non-manifoldness is caused by CSG operations on coincident
> > > faces. There needs to be some slight overlap when doing difference and
> > > subtraction and also often with union as well.
> >
> > I have done that, the overlap is 1mm.
> >
> > > I use netfabb studio to highlight the errors if I am stuck and can
> > > generally work out the problem for there. I don't get many but I do
> > > have
> >
> > a
> >
> > > small bodge factor all over the place. Look for eta in the source for
> > > Mendel90 for example. It is defined as 0.01 and gets added or
> > > subtracted
> >
> > to
> >
> > > make sure faces are not exactly coincident.
> >
> > Can netfabb read scad files?  As far as I can see it only reads STL and
> > this
> > happens when I try to export the STL.
> >
> > David
> >
> > > On 11 May 2012 14:54, David Goodenough <
> >
> > [hidden email]>
> >
> > > wrote:
> > > > I have a design which apparently works when you press F5, and also
> > > > when
> > >
> > > you
> > >
> > > > press F6, but when you come to export it as STL, then it fails.
> > > >
> > > > I have reduced it a bit, and I have two major components each of
> > > > which
> >
> > on
> >
> > > > their own will F5, F6, and export as STL, but when you take their
> > >
> > > difference
> > >
> > > > it fails.
> > > >
> > > > It would be really useful to have some visual clue to show where the
> > >
> > > problem
> > >
> > > > might be.  The nearest comes when I do an F5 on the difference, and
> > > > an odd green face appears (for no good reason that I can see) but
> > > > only
> >
> > from
> >
> > > > some viewpoints - move the viewpoint and it vanishes.
> > > >
> > > > Apart from brute force, does anyone have any good debuging techniques
> >
> > in
> >
> > > the
> > >
> > > > absense of visual clues?
> > > >
> > > > David
> > > > _______________________________________________
> > > > OpenSCAD mailing list
> > > > [hidden email]
> > > > http://rocklinux.net/mailman/listinfo/openscad
> >
> > _______________________________________________
> > OpenSCAD mailing list
> > [hidden email]
> > http://rocklinux.net/mailman/listinfo/openscad


Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

Alan Cox
In reply to this post by David Goodenough
On Fri, 11 May 2012 14:54:46 +0100
David Goodenough <[hidden email]> wrote:

> I have a design which apparently works when you press F5, and also when you
> press F6, but when you come to export it as STL, then it fails.
>
> I have reduced it a bit, and I have two major components each of which on
> their own will F5, F6, and export as STL, but when you take their difference
> it fails.

I've hit this repeatedly with differencing two objects. I now extend the
any holes that should go through the outer object further out which seems
to do the trick.

If you offset them a tiny amount (eg 0.2mm) then look at it you can often
see which holes have vanished as a guide to where the problem is likely
to be

You may also find scaling the inside to 100.01% fixes it



Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

tjhowse
I have a variable called z_fight_fudge, or zff, set to 0.005. I liberally throw this around to get rid of coincident faces.

On 12 May 2012 02:10, Alan Cox <[hidden email]> wrote:
On Fri, 11 May 2012 14:54:46 +0100
David Goodenough <[hidden email]> wrote:

> I have a design which apparently works when you press F5, and also when you
> press F6, but when you come to export it as STL, then it fails.
>
> I have reduced it a bit, and I have two major components each of which on
> their own will F5, F6, and export as STL, but when you take their difference
> it fails.

I've hit this repeatedly with differencing two objects. I now extend the
any holes that should go through the outer object further out which seems
to do the trick.

If you offset them a tiny amount (eg 0.2mm) then look at it you can often
see which holes have vanished as a guide to where the problem is likely
to be

You may also find scaling the inside to 100.01% fixes it


_______________________________________________
OpenSCAD mailing list
[hidden email]
http://rocklinux.net/mailman/listinfo/openscad

Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

nophead
Z buffer fighting when you do F5 is different from non-manifold with F6, but they often go together. You can get Z buffer artefacts when unrelated faces are close to each other and faces that have been removed, so are not actually there in the finished object. And you can get non-manifold objects that don't show up on F5.

When I did the full assembly of Mendel90 I had to put render statements around all the individual components to get rid of  Z buffer artefacts and give reasonable drawing speed.

On 12 May 2012 03:13, tjhowse <[hidden email]> wrote:
I have a variable called z_fight_fudge, or zff, set to 0.005. I liberally throw this around to get rid of coincident faces.


On 12 May 2012 02:10, Alan Cox <[hidden email]> wrote:
On Fri, 11 May 2012 14:54:46 +0100
David Goodenough <[hidden email]> wrote:

> I have a design which apparently works when you press F5, and also when you
> press F6, but when you come to export it as STL, then it fails.
>
> I have reduced it a bit, and I have two major components each of which on
> their own will F5, F6, and export as STL, but when you take their difference
> it fails.

I've hit this repeatedly with differencing two objects. I now extend the
any holes that should go through the outer object further out which seems
to do the trick.

If you offset them a tiny amount (eg 0.2mm) then look at it you can often
see which holes have vanished as a guide to where the problem is likely
to be

You may also find scaling the inside to 100.01% fixes it


_______________________________________________
OpenSCAD mailing list
[hidden email]
http://rocklinux.net/mailman/listinfo/openscad


_______________________________________________
OpenSCAD mailing list
[hidden email]
http://rocklinux.net/mailman/listinfo/openscad


Reply | Threaded
Open this post in threaded view
|

Re: is there a way to highlight non-manifoldness?

Whosawhatsis
It should be noted that the 2D subsystem handles this stuff much better and does not need fudge factors. I find it's best to do as much as possible in 2D, then extrude. This also helps immensely in avoiding hitting the "too many entities" (or whatever it's called, I've managed to avoid it for quite a while) limit.

On Saturday, May 12, 2012 at 2:35 AM, nop head wrote:

Z buffer fighting when you do F5 is different from non-manifold with F6, but they often go together. You can get Z buffer artefacts when unrelated faces are close to each other and faces that have been removed, so are not actually there in the finished object. And you can get non-manifold objects that don't show up on F5.

When I did the full assembly of Mendel90 I had to put render statements around all the individual components to get rid of  Z buffer artefacts and give reasonable drawing speed.

On 12 May 2012 03:13, tjhowse <[hidden email]> wrote:
I have a variable called z_fight_fudge, or zff, set to 0.005. I liberally throw this around to get rid of coincident faces.


On 12 May 2012 02:10, Alan Cox <[hidden email]> wrote:
On Fri, 11 May 2012 14:54:46 +0100
David Goodenough <[hidden email]> wrote:

> I have a design which apparently works when you press F5, and also when you
> press F6, but when you come to export it as STL, then it fails.
>
> I have reduced it a bit, and I have two major components each of which on
> their own will F5, F6, and export as STL, but when you take their difference
> it fails.

I've hit this repeatedly with differencing two objects. I now extend the
any holes that should go through the outer object further out which seems
to do the trick.

If you offset them a tiny amount (eg 0.2mm) then look at it you can often
see which holes have vanished as a guide to where the problem is likely
to be

You may also find scaling the inside to 100.01% fixes it


_______________________________________________
OpenSCAD mailing list
[hidden email]
http://rocklinux.net/mailman/listinfo/openscad


_______________________________________________
OpenSCAD mailing list
[hidden email]
http://rocklinux.net/mailman/listinfo/openscad


_______________________________________________
OpenSCAD mailing list