Mirror, mirror on the . . . um . . . screen

[Anonymous]

Been mirroring sketches for some time with no trouble. Handy because I usually work on one half of a symmetrical drawing to get the bugs out, and then I mirror it.

Yesterday I mirrored half a surface of a musical instrument arching. It appeared to go well, and when I was done all I had to do was stitch the two halves together in preparation for thickening it. Then things got odd. After the mirror, the two halves were displayed in F360's default steel color. After the stitching, the original half stayed steel colored, while the mirrored half took on a nice brassy color.

When I thickened the surface, one half thickened in the plus Z direction, and the other in the minus Z direction. This seems to make any further operations, such as trimming, impossible. Is this a bug? Anyone know how to squash it?

[wmhazzard]

Can you post a f3d file of your problem. I could not replicate your result. 

[davebYYPCU]

In Surface work you colour differences is like inside / outside, (Normal)

Use Modify > Reverse Normal, so that both are of the same orientation, 

It’s a 50 / 50 thing.

Might help...

[TheCADWhisperer]

---

@Anonymous wrote:

...Anyone know how to squash it?

---

Reverse Normal!

[Anonymous]

Here's the latest. Used flip normal, and the colors reversed so now the left side is steel and the right side is brass. After a few tries (what is it about human nature that makes us try something that doesn't work one time so that it doesn't work five times?😂) watching the colors flip back and forth, I decided it might be helpful if I unstitched the halves, which I did. Of course, when they were no longer stitched, both had the same color showing. But, of course, I flipped the offending side anyway, stitched the two back together,

[Anonymous]

Whole message this time!

Here's the latest. Used "flip normal", and the colors reversed so now the left side is steel and the right side is brass. After a few tries (what is it about human nature that makes us try something that doesn't work one time so that it doesn't work five times?)  watching the colors flip back and forth, I decided it might be helpful if I unstitched the halves, which I did. When they were no longer stitched, both had the same color showing. But, of course, I flipped the offending side anyway, stitched the two back together, and got exactly the same mismatch as before.

Kinda curious . . .

[g-andresen]

Hi,

then follow the request of @wmhazzard and share the design to verify it.

günther

[laughingcreek]

There is almost certainly a problem in your model.  Posting your .f3d file will get you better help.

[Anonymous]

Sorry . . .

Trying to fit this in between clients. I'm hoping to get the solution soon because all I would need to do after that is to convert it to a solid, after which I could get it into the CAM environment to see what incredible events await me there! ☹️

[laughingcreek]

The t-spline edge down the middle crosses back and forth over the centerline.  You can see the effect this causes if you zoom in close and look at the intersection of the 2 surfaces.  in some places they over lap, and in some places there is a gap between the 2.

You can kludge this together by changing the inputs for the first stitch feature in your timeline.  Set the tolerance all the way up and it will stitch it together the way you want.

That my be good enough for you, but there are other problems.  the surface is kinda lumpy , and down the centerline the arch dips into a valley, which is going to be a b*tch to sand out if you CNC this. (done that).

We can talk about further refinements if your interested.  Fwiw, I've attached a quick example of how I approach these types of things.  I think I posted it in a previous post of your also?  adding star points and edge loops that follow the contour will give you more control over the re-curve,  using fewer faces will make help in making the curved surface more fair, and using mirror within the t-spline body will help with the transition across the center line.

[TrippyLighting]

One would assume that you re trying to establish at least tangency (G1)  if not curvature continuity (G2) across the mirrored surfaces.

That will NOT ever happen with the methods you've chosen!

Lofting with the tools in the surface tab requires a different sketch layout than the one you've chosen so that's pretty much out. Just the outlines in your sketches already  have way to many individual segments to yield smooth curvature and thy are all not constrained to be tangent. For lofting with the surface tools these sketches are hopelessly useless.

You then attempted to create this in the T-Spline environment and were successful in that you were able to create "some" surface. However, if you look at that surface in box view mode, just that already will tell you that this is not going to generate a smooth surface.

As I show in the screencast you can use the flatten command to pull the vertices that are not perfectly aligned toward with center plane to that center plane and then you can mirror-duplicate the surface.

But even without the analysis tools it can already be seen that there is going to be a dip in the center of the surface.

The Zebra analysis shows the "wobbliness" of the T-Spine surface also clearly.

[Anonymous]

The last few responses have been humbling, to say the least. While I am applying bandages to my wounds, 😥 I will reflect on the consequences of not knowing all the things one doesn't know.

So allow me to put this as plainly as I can; I need help. I need a mentor or mentors. I need to push through this undertaking, and I am under some time constraints for reasons that we needn't go into here.

It has been difficult to find people who have a grasp of what I'm trying to achieve as a new user. Following the advice of the well-meaning YouTube "tutorials," I can only say that the usual idea to take splines or faces and push and tug them around "until things look right" isn't going to cut for me. Whatever the pain, it will be better for me to get get it right out of the gate rather than spin my wheels with endless fixes and repairs.

It sounds like I should start with a complete sketch rather than my usual half-sketch-and-mirror approach, unless there is a way to draw mirror lines as I go. Then the arching guidelines should be tangent at the centerline, just for starters

Peter, your experience is daunting. You are way deeper into this than I even knew one could go, and it'll take me a while just to figure out what you are doing. Many thanks for taking the time to make the screen cast.

OK, time for a bite to eat, and then maybe I will practice my bass for an hour! 😀

[TrippyLighting]

This isn't at all a beginner project so its not a surprise that you are encountering difficulties.

If you are trying to create a real violin with a highly glossy surface we better make sure that the surface you are milling is as smooth as it can be. This is in essence a class A surfacing problem.

Please do explain what your ultimate goal is with this. These sketches were done with a goal in mind, that much is clear. That grid was done with a purpose. It reminds me of techniques I used as a kid when trying to copy existing drawings I liked. I created a grid and where the contour of an object crossed a grid line I made a point. Later I connected the dots.

Are you trying to create an exact replica of an existing instrument ?

You mentioned CNC. What is it you are to mill with the CNC. ?

The large curved top ( and bottom) surfaces splines with as few ams possible spline points are your friend. 

I am experimenting:

[TheCADWhisperer]

I am curious - did you see the video that I posted for you at the end of this previous discussion?

https://forums.autodesk.com/t5/fusion-360-design-validate/how-would-you-constrain-this-drawing/m-p/9...

[Anonymous]

It's amazing how easy it is to see something once someone else has pointed it out to you. 😁 Granted the surface is lumpy, but it gets cut thick in order to allow hand-corrected operations for minor variations later on. Not he ideal way, for sure, but at least I have the satisfaction of knowing that this latest attempt was far, far better than my previous effort! If you can get me up to speed on star points and edge loops it'll help me better understand your excellent suggestions. Their function and how to use them are new to me.

It would help enormously if Fusion allowed the user to snap points on the T-spline planes to points or lines on the drawing. That would correct the center line problem. I must say that I was confused to learn that stitching the halves of the faces left overlaps and gaps between them. I saw these, too, when I zoomed in very close (actually much closer than I would have thought to do). Would using "align" have addressed that? Also, I did not see the attachment you posted.

Thanks for your comments. I really appreciate them.

[TrippyLighting]

---

@Anonymous wrote:

It would help enormously if Fusion allowed the user to snap points on the T-spline planes to points or lines on the drawing. That would correct the center line problem.

---

Not true!

I also already answered in my screencast how that T-Spline surface can be mirrored.

[Anonymous]

Hi, Peter--

I'll do my best to answer your wonderful response in less than 10,00 words!😊

For starters, my goal would be to get as far as I have presently struggled, only to get it right. I'm approaching this with the idea of stages. The first stage would give me a " free plate" (that's what luthiers call the top or back of a violin when they are not attached to the rib assembly). The plate could be cut on both sides by a CNC machine.

The next step would be routing the purfling groove (contains the decorative banding nearer the edges of bowed string instruments), cutting the recurve, (the point where the arching reverses, dipping slightly below the plane of the sketch and then rising to a given distance from the outside edge of the plate), and then cutting the outline. This would require three tool changes, and I don't think I'm anywhere near ready for that yet.

Philosophically, I would look at the entire process and whatever work that the machine could do better or faster than the hand worker would be given to the machine, and whatever work can done better and faster by hand would be left to me. It's a process I hope will give me a base from which I could make improvements later.

The instruments are members of the string family that are either newly developed or that disappeared from use several centuries ago. While they are every bit violins, they are not made commercially. As to the level of fit and finish, I can accept less than shiny surfaces at first since I've already assumed that at least some handwork will be required no matter how well a machine can be programmed to do it.

Of course, none of this happens until I get to a certain point with Fusion 360. There's an "aha!" moment somewhere out there ahead of me, and I think things could go easier once I experience it. Especially with the help of folks like you and the others here.

And just to end on a humorous note, the "360" label on the drawing does not refer to Fusion 360, but rather to the length of the body in millimeters. Biggest family member, the sub-contra, is over 1200 mm, so you can see the advantages of starting small.

--Bob

[wmhazzard]

I was wondering where you got your data from to design the violin? It looks like you have very specific measurements and maybe there is a better way to draw it in Fusion. There is a way to import a CSV file from an Excel spread sheet which would give you the curves that you want and you could then loft between the curves to make the surface.  Just throwing that out there as another possible way to better draw the violin. 

Bill

[TrippyLighting]

---

@wmhazzard wrote:

I was wondering where you got your data from to design the violin? It looks like you have very specific measurements and maybe there is a better way to draw it in Fusion.

---

That is also what I am trying to get to the ground of!

[TrippyLighting]

---

@wmhazzard wrote:

 There is a way to import a CSV file from an Excel spread sheet which would give you the curves that you want and you could then loft between the curves to make the surface. 

---

If you are talking g about lines and arc, that would work. For splines, however, that would not work at all.