Bspline strange behaviour

[carlopav]

Hello! playing with bspline to allow edit tool preview, i noticed a strange behaviour when the spline is closed:

Cattura.JPG (23.39 KiB) Viewed 773 times

it seems strange to me that a closed spline (red one) generated from rectangle's midpoints, it's not simmetrical.
I highlighted the spline points in every attempt to be clearer. Maybe it's just my low math skills

Code: Select all

OS: Windows 7
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.19.16687 (Git)
Build type: Release
Branch: master
Hash: ff1fb11af15a7c5f7f0ba3ef79588e128f397c2c
Python version: 3.6.6
Qt version: 5.6.2
Coin version: 4.0.0a
OCC version: 7.3.0
Locale: Italian/Italy (it_IT)

[renatorivo]

This also:

Cattura-b-spline.PNG (20.48 KiB) Viewed 720 times

B-spline with SolidWorks

splineSolidW.PNG (54.24 KiB) Viewed 720 times

[bitacovir]

This also:
Cattura-b-spline.PNG

Here I think is because the last points are not curves.

[yorik]

I think these b-splines as created "as is" by opencascade I think (might be wrong there, I'll have a look), in that case there isn't much that we can do on our side...

[chrisb]

All these B-splines seem to be correct. The problem is that we cannot add additional constraints, especially tangents.

[carlopav]

I think open bspline works quite good, and the example Renato added is correct. (Imagine a thin wooden board that have to go around nails, represented by knots: second knot will never have an horizontal tangent if the first one is free to rotate). How does it behave in SW? A CAD ask you to set first and lastpoint tangent, but then it leads to awful representation when u move them, and it's hard to change them also...
It looks more what crisb pointed out: the problem occurs when the spline is closed because it seems to me that the tangent of the first and last point is not correctly calculated (last point takes the tangent of the first point as is, without modifying it).
Of course if Yorik is right we have to keep it like this... At the end it's not a major issue and we can always use bezcurves