Curvature Query

[Anonymous]

Hi,

Many thanks for taking the time to view my query / forum post.  I initially posted this query in the Collaborate forum by mistake (should hopefully be removed now).

BACKGROUND:
I am trying to replicate (to scale) a real cylinderical-ish body in Fusion360, with the hope of being able to 3D print an accurate copy.

Thus far I have taken profiles / formers using cardboard of said object, transferred these to Fusion, checked 3d printed prints of the profiles against the object, followed by a lot of iteration until the profiles seemed to match.  I have tried photogrammetry, but had little success (though this is an avenue I may revisit).  When I loft between all profiles (Loft 1), I notice there are areas with high curvature around the mid section (Loft 1_XY profiles_curvature.jpg).

Initially, I thought this might be because the loft function was having difficulty interpolating between the profiles (those in the YX plane).  To address this, I tried adding guide rails at 0, +/-30, +/-60 & +/-90 degree planes (rotated about the Z axis). The guide rails comprised splines connecting to points that coincide with where the XY profiles intersected the sketch plane; I was hoping the splines would smooth out the loft.  Repeating the loft using said guide rails and the profiles at z_0mm to z_+276mm only, did seem to improve things (Loft 2_guide rails only_curvature.jpg).  It did however result in two localised areas of high curvature, concentrated along the +/-30 degree guide rails, mid-section.

I then tried a further loft (Loft 3) using all XY profiles and guide rails. This approach seemed to reduce the high curvature seen in Loft 2, but did not eradicate it / reduce it to less than that seen in Loft 1.

QUESTION:
Is it possible to reduce / eliminate this curvature or will it always exist with the geometry I am trying to copy? Is there a different approach (or function) I could try that might yield better results?

Any help / advice will be much appreciated.  I have had a look online for anything, but I couldn't find much regarding curvature issues / fixes.

Apologies for the length of the post.

Kind regards,

LyleMann

[Anonymous]

Hi,

Many thanks for taking the time to view my query / forum post.

BACKGROUND:
I am trying to replicate (to scale) a real cylinderical-ish body in Fusion360, with the hope of being able to 3D print an accurate copy.

Thus far I have taken profiles / formers using cardboard of said object, transferred these to Fusion, checked 3d printed prints of the profiles against the object, followed by a lot of iteration until the profiles seemed to match.  I have tried photogrammetry, but had little success (though this is an avenue I may revisit).  When I loft between all profiles (Loft 1), I notice there are areas with high curvature around the mid section (Loft 1_XY profiles_curvature.jpg).

Initially, I thought this might be because the loft function was having difficulty interpolating between the profiles (those in the YX plane).  To address this, I tried adding guide rails at 0, +/-30, +/-60 & +/-90 degree planes (rotated about the Z axis). The guide rails comprised splines connecting to points that coincide with where the XY profiles intersected the sketch plane; I was hoping the splines would smooth out the loft.  Repeating the loft using said guide rails and the profiles at z_0mm to z_+276mm only, did seem to improve things (Loft 2_guide rails only_curvature.jpg).  It did however result in two localised areas of high curvature, concentrated along the +/-30 degree guide rails, mid-section.

I then tried a further loft (Loft 3) using all XY profiles and guide rails. This approach seemed to reduce the high curvature seen in Loft 2, but did not eradicate it / reduce it to less than that seen in Loft 1.

QUESTION:
Is it possible to reduce / eliminate this curvature or will it always exist with the geometry I am trying to copy? Is there a different approach (or function) I could try that might yield better results?

Any help / advice will be much appreciated.  I have had a look online for anything, but I couldn't find much regarding curvature issues / fixes.

Apologies for the length of the post.

Kind regards,

LyleMann

[TheCADWhisperer]

Did you intend for Profile_z_0mm to be asymmetrical?

Do you have any real world dimensions?

[Anonymous]

Hi TheCADWhisper, thanks for replying to my query.

Yes, the profile at z_0mm is meant to asymmetrical (as in, it's not meant to be a half-circle). The object itself is symmetrical, but only about the ZY plane - I thought it would be quicker / easier to model half of it then mirror about said plane.

Kind regards,

lyle.s.mann

[laughingcreek]

First things first. get a handle on your input curves.  messy input curves = messy surfaces.

here's a pic with curvature combs turned on-

second, when it comes to using splines for lofting, less is more.  you want to use as few points and cross sections  as possible to define your lofts.

[Anonymous]

Hi laughingcreek, thanks for taking the time to reply to my query.

Crikey, I see what you mean, the profiles are very messy. How best would I go about correcting this / smoothing the profiles? Based on your feedback, would the best approach be to 1) re-draw / approximate the existing profiles, using as fewer spline points as possible 2) reduce the number of profiles used when lofting, balanced against the shape I am trying to approximate? Is there a function / method in Fusion 360 that automates (or aids with) step 1 or is this manual process?

[laughingcreek · Accepted solution]

correct on both accounts.

no automated process in fusion to do it.  manual process.

[Anonymous]

Understood, thanks laughingcreek, your help is much appreciated.

Kind regards, lyle.s.mann

[jeff_strater]

I think this is a duplicate of this other post:  curvature-query .  Please refer to that thread.

[TheCADWhisperer]

---

@Anonymous wrote:
Hi TheCADWhisper, thanks for replying to my query.
 (as in, it's not meant to be a half-circle). 

---

That is not at all what I meant.

I could see that it was not a half circle.

[TheCADWhisperer]

https://autode.sk/3cqccH6

[Anonymous]

Ah, apologies for the misunderstanding, yes I intended to mirror about the ZY plane.

Thank you very much for taking the trouble to post a screencast, it is really appreciated. I wanted to try out your suggestion, and laughingcreeks', before replying, hence the delay.

I've attached two screen shots (and the .f3d file), one that shows half of the body modelled with smoother spline profiles, the second using a single spline per profile.  Both resulted in notably better lofts than my attempts.

I haven't quite eliminated the curvature entirely (assuming it is possible to do so, without deviating from the original shape too much?), which I suspect is because I didn't match the profiles against the body properly - I am currently rechecking them.

Many thanks again for your assistance,

lyle.s.mann

[laughingcreek]

as long as your experimenting, you might want to go to the extreme end of "less is more". ie start with less and work toward more as needed, instead of the other way around.

I attached a few more approaches.  (added to your model as component1, component3, and component 4.

component1 uses solid modeling.  you'll note it isn't tangent across the mirror line in places.  this is because the widest point in the x direction and the widest point in the y direction falls at a different point along the z axis.  profile 2 is roughly at the widest point in the y direction.  the loft gets pulled out of tangent slightly along the rails because of the arrangement. (I tried reversing the profiles and the rails so I could use the direction condition on the profile to fix this.  doesn't work.  I had a conversation with AD a while back about this.  they said once you add rails, the direction condition is no longer considered.  don't know why.  Don't know if they consider it a bug (I do) or just the way it should work.  I've yet to find a situation where the direction condition was useful, mostly b/c of the inability to use rails)

component3 uses the same input curves.  but instead of solid modeling, I'm using surfaces.  I pulled helper surfaces off the rails, and when I did the loft I used the edges instead of the sketches for the rails, and set the condition to tangent.  the result is much better. (it's tangent across the mirror line.)

component4 is only marginally better curvature wise.  I replaced profile 2 with one set at an angle.  In this case I literally drug the construction line used to define the sketch plane around   till the curvature looked the best.  there's probably better ways to figure out where this profile should be, but fusion doesn't really provide the tools to help with that.

enjoy

[Anonymous]

It is, apologies.

[shahriarsifat1802164]

Hi,
I think this will remain if you copy that particular geometry.
Thank you.

[shahriarsifat1802164]

Again my recommendation would be to try to activate the curvature comb tool.

[shahriarsifat1802164]

Reducing the spline would be an option also.

[shahriarsifat1802164]

Lastly, all you have to do is manual. Because fusion allows you to perform your works manually.
Thank you.

[Anonymous]

Apologies for the delay in replying, I wanted to (try and) thorough understand your post (and inspect the model) before replying.  I also apologise in advance if any of the following queries seem noobish (for lack of a better word) - I'm still very much on the F360 learning curve.

component1:
When you say 'you'll note it isn't tangent across the mirror line in places' is this where the zebra lines don't run parallel-ish with the ZY plane (or bulge) around profile 2, when viewed from the 'TOP'?

Is '...the loft gets pulled out of tangent slightly along the rails' are you referring to the same thing here?

My rudimentary understanding of the loft function's Direction option / condition is that the body is tangent to the start & end profiles (i.e. profiles 1 and 3) only?

'tried reversing the profiles...' do you mean you altered the order that the profiles were specified in i.e. after trying lofting from e.g. Profile 1, 2 then 3, you then tried selecting Profile 3, 2 then 1?

component3:
Again, apologies for the beginner-level question...
By extruding the sketched rails, when specifying the loft, the connection to the edge of the rails can be specified as tangent; unlike when in 'SOLID' mode, there is no option to specify tangency along the loft guide rails, and thus, the body's outer surface could be tangent or non-tangent (at points along the ZY &/or ZX planes) once lofted, as seen in component1?

component4:
Is the idea here to try and get the body to intersect Profile 2 at right angles?
Would the ideal Profile 2 be one that is on a curved sketch plane, such that the lofted body intersects this profile at 90° at every point?

Many thanks again for your help - for what it's worth I'm learning a lot from this thread.

Kind regards,

lyle.s.mann

[Anonymous]

Hello,

Thanks for taking the time to reply.  Apologies, I should have added a hyperlink to my earlier post: https://forums.autodesk.com/t5/fusion-360-design-validate/curvature-query/m-p/10048514/highlight/fal...

I have tried adjusting the splines / approximating them with fewer control points (as you suggest), which as improved the curvature somewhat.  But I suspect that what I am trying to achieve might not be possible.

Kind regards,

lyle.s.mann