Fillet problems

[Anonymous]

I'm having trouble filleting the contoured edges of the attached dwg.  I'm getting the error, "could not find acceptable sequence of capping faces to trim blend face."  The workaround that was working was to slice the part into smaller sections to where the fillet would work, then union them back together.  It no longer works at the bridge of the nose.  I kept having to slice very thin, but hit a wall.  Suggestions?

[JDMather]

That geometry is very rough - you have 1778 faces, 67 of which do not pass quality check in Autodesk Inventor.

You have a lot of planar faceted faces where I would expect smooth flowing curves in a product like this.

Is it too late to start over?

Also the -dwgunits command indicates that your current document units are inches and dist indicates that your design is approximately 1,540 inches wide.

And the design is skewed it two planes?

It must be very difficult to work on this design.

[Anonymous]

I'm not sure what the quality check means.  Also, the uploaded dwg is just a copy of part of the actual dwg I'm working on.  I just copied the part I needed help on into another dwg.  The units are in mm in the original.  The planar faceted faces I think you're referring to are where I had to slice, fillet, then union; causing that slightly step-down look.  It is 10x larger than what I'm working on.  Was hoping it would work at the larger size, then I could scale back 10x smaller.  Even if I start over, I believe I'll have the same issue, unless there's a better method prior to the current state.  Thanks.

[JDMather]

---

@Anonymous wrote:

1. I'm not sure what the quality check means.

 2. Also, the uploaded dwg is just a copy of part of the actual dwg I'm working on.

 3. I just copied the part I needed help on into another dwg.  The units are in mm in the original.  

4.The planar faceted faces I think you're referring to are where I had to slice,

5.  Was hoping it would work at the larger size, then I could scale back 10x smaller.

6. Even if I start over, I believe I'll have the same issue, unless there's a better method prior to the current state.  Thanks.

---

1. Don't worry about that - this is not in AutoCAD, it just means that I checked the geometry in another CAD program.  But what you do need to worry about is that the other CAD program reported that the geometry is poor quality.

2. Did you create the original geometry or did you get it from someone else?  Did you preserve the construction geometry in the original (delobj=0)?  Standard practice for complex geometric designs.

3. Did you start from metric template or set -dwgunits before copying the geometry?

4. I don't think so.  I think the planar faceted faces are due to starting with line profiles rather than spline (or tangent arcs and lines).  In any case hopefully you have saved this file before you attempted this ill fated work-around and can attach the clean file here.

5. First time I have heard of this logic.  Has it worked for you on other projects?  Perhaps if it was very very tiny geometry or very very large geometry (100,000cm bounding box math) scaling would make a difference.

6. I am quite certain there is a better technique.  What was your very first step in this design?  Using someone else's work?  Be specific and attach dwg and/or image files to illustrate your description of step.

________

I would start by setting -dwgunits

I would set delobj=0

I would set facetres=10

I would set viewres=2000 or greater

I would create several layers for control of specific structures of geometry, ex. Construction Profiles, Solid Bodies....

Probably a bunch of other preliminary steps that I have forgotten because they are in my standard template and I don't think about them.

Anyhow - getting your starting step should help guide the rest of the design.

Edit:  And my design would most definitely be symmetrical about the origin.

[wispoxy]

@Anonymous I'd start over like @JDMather questioned if you want your product smooth. I looked into the cad file and one issue just leads to another. Maybe copy out the base outline flat and restart, preserve what you can. I'll play with it more though.

[Anonymous]

1. Don't worry about that - this is not in AutoCAD, it just means that I checked the geometry in another CAD program.  But what you do need to worry about is that the other CAD program reported that the geometry is poor quality.

2. Did you create the original geometry or did you get it from someone else?  Did you preserve the construction geometry in the original (delobj=0)?  Standard practice for complex geometric designs.

3. Did you start from metric template or set -dwgunits before copying the geometry?

4. I don't think so.  I think the planar faceted faces are due to starting with line profiles rather than spline (or tangent arcs and lines).  In any case hopefully you have saved this file before you attempted this ill fated work-around and can attach the clean file here.

5. First time I have heard of this logic.  Has it worked for you on other projects?  Perhaps if it was very very tiny geometry or very very large geometry (100,000cm bounding box math) scaling would make a difference.

6. I am quite certain there is a better technique.  What was your very first step in this design?  Using someone else's work?  Be specific and attach dwg and/or image files to illustrate your description of step.

_____

1.  How  does geometry become of "poor quality?"

2.  I created it from scratch.  Haven't heard of using that command, but will do in the future.  

3. Started at metric. The one I uploaded was inadvertently copied into a dwg with inches.

4. I'm not sure what faceted faces you're referring to.

5. I saw this suggestion in a few forum threads, so I tried it.

6. I used splines and arcs to create the front view of the glasses and a top view of the glasses.  Then joined with pedit, extruded, convtosurface, surftrim to remove excess, convtosolid, filletedge and had to slice  the part into segments to get it the fillets to work.

I tried these suggestions to no avail, but I'm guessing it would only matter if implemented at the start:

I would start by setting -dwgunits

I would set delobj=0

I would set facetres=10

I would set viewres=2000 or greater

[JDMather]

---

@Anonymous wrote:

6.
a) I used splines and arcs to create the front view of the glasses and a top view of the glasses.  

b)Then joined with pedit, extruded,

c)convtosurface, surftrim to remove excess,

d)convtosolid, filletedge

e)and had to slice  the part into segments to get it the fillets to work.

---

Do you still have a copy of the file at a state before you did step e)

a) If you set delobj=0 and create layers for your construction geometry - you never loose any work and do not have to start over from the beginning when you run into trouble.

b) If you create copies at critical changes in geometry and preserve these file revisions or hidden layers - you never loose any work and do not have to start over from the beginning when you run into trouble.

c) See b)

d) See b)

e) ?

If you do not follow these steps - then you have no choice but to start over. (and I recommend that you follow these steps - delobj=0, layers and back-up revision files are (or should be) your best friends)

I often tell people to start over and then balk - ...but I already have two weeks into this...

So they continue to fight it...

The night before the project is due they finally resign to starting over and find that in a couple of hours they finish.

When you start over - I recommend that you attach each step from a) to c) here as you go along.

That way you can benefit from the extensive combined experience of this community.

[Anonymous]

Ok, so I back-tracked and started from the polylines to extrude and used only solids with intersect instead of surfaces with surftrim.  And I'm back to being unable to fillet.  You'll see the nose-pad was created with mesh and convtosolid, but that's irrelevant for now.  See the attached pic of just one section that won't fillet.  Also attached the dwg I'm working on.  You'll see it's messy.  I was copying as I went along to have a history of objects to use.  I just don't understand why some areas of the object fail while filleting.

[JDMather]

You went wayyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy too far without asking any questions - now you will have to start over from scratch.

I instructed you to set your document units with -dwgunits command.

Your glasses are around 133 inches across.

But that is the least of your concerns.

You have wayyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy too many nodes in your profile.

If I explode the curve I see that it is made up of thousands of tiny lines.

You will never derive good geometry from this.

You should have only a handful of arcs and lines that define the profile.

There are other issues besides this, but this is the foundational issue and there is no point in continuing beyond this without a robust foundation.

[parkr4st]

I agree with the advice.  Start afresh.

your frames have many oddities in the line work.

Attached dwg, erased the inner line and offset 2.6 units in from the outer line, and then erased the outer line and offset from the new inner line out 2.6 units.  most of the oddities are gone and need tweaking.  Zoom way in and that is obvious.

I put arcs in the upper lens openings and trimmed and joined.

extruded the solids and subtracted

my work is inside the polygon

looks neater but still is not 100% drawn correct IMHO.  This should lead to a smoother edge which the fillet will behave better on.

dave

[JDMather]

---

@Anonymous wrote:

...  I just don't understand why some areas of the object fail while filleting.

---

Q1.  Do you have any desire to learn how to model this correctly?

[Anonymous]

1.You went wayyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy too far without asking any questions - now you will have to start over from scratch.

        I didn't know I needed to ask questions, all seemed to be going fine until I was unable to fillet.

2. I instructed you to set your document units with -dwgunits command.

Your glasses are around 133 inches across.

       I have it set to mm, so 133 mm.  Is there a difference in setting units in the GUI vs with dwgunits?

3. You have wayyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy too many nodes in your profile.

If I explode the curve I see that it is made up of thousands of tiny lines.

You will never derive good geometry from this.

       I didn't know that was a problem.  Obviously it is.  I just did it again with only about 25 splines/arcs to the same fillet problem.

[Anonymous]

Q1.  Do you have any desire to learn how to model this correctly?

        Yes.

[JDMather]

To model this correctly will require very precise set of steps.

Units command and -dwgunits (with the preceding - sign, exactly as written) are two different commands.

Before we begin, what is the source of your original geometry? 

[Anonymous]

I took measurements of my glasses with calipers, took pics and uploaded them to autocad.  Then scaled them to size and traced.

[JDMather]

---

@Anonymous wrote:

I took measurements of my glasses with calipers, took pics and uploaded them to autocad.  Then scaled them to size and traced.

---

That is the correct procedure, but this time we will take a precise route with the "tracing" step.

Start a new part file using the acadiso3d.dwt template.

Verify that the document units are mm using the -dwgunits command.

Set delobj=0

Set viewres=2000 or greater

Set Facetres=10

Create a layer named something like Imported Image.

Place the image on that layer.

Scale the image.

Center the nose bridge of the image at 0,0,0.

Attach the *.dwg file here for next set of steps.