Curves workbench

[Vincent B]

Lofted output there aren't any error with theses parameters.
why?

[drmacro]

I switched the posted file to lofted, solid, it still failed.

But, if I reduce the sample to 40 it works. 80 samples fail. 60 samples fails. 45 samples...crash.

[drmacro]

I'm wondering if this is one of those situations where the geometry check just isn't up to the task.

I tinkered more with this, changing some shapes. It produces a nice tube. But, essentially fails the geom check with almost all settings.

Snip macro screenshot-d5eedd.png (113.94 KiB) Viewed 1939 times

But if I mesh it and use Blender to check if it's manifold it reports no issues. It is a really messy mesh...

BlenderMeash.png (82.09 KiB) Viewed 1939 times

[drmacro]

Here the Gordon test ok with geometry check.

Snip macro screenshot-480701.png (159.62 KiB) Viewed 1885 times

If I now create an offset, the edges created by the offset fail geom check.

Snip macro screenshot-8b903d.png (159.1 KiB) Viewed 1885 times

I have tried various settings in the Shape Approximation properties of the JoinCurves to no avail.

I'm just trying to figure out how to create valid solids from things in the Curves workbench. It seems rather hit or miss, and I want to understand if it's me, if I need more understanding of some factors, etc.

(note: deleted the offset to get the file size down.)

[Vincent B]

reduce Max Ctrl Pts to 20, make a simpler gordon geometry. Then there are not error.

[drmacro]

reduce Max Ctrl Pts to 20, make a simpler gordon geometry. Then there are not error.

I'm trying to get some sort of rule of thumb or general understanding that could be communicated to others or in a wiki.

I have tried a series of reducing points (in this gordon and samples in pipeshells). There seems to be no methodical way to define how to make solids that consistently pass geometry check. Trial and error appears to work sometimes, but not others.

Yet, in most cases, a mesh of the solid, though messy, appears to be valid.

[drmacro]

Did the behaviour of the GeomInfo change?

I remember it staying on until toggled off.

Now any action that results in nothing being selected hides the display. (Selecting something unhides.)

And, a question: would it be possible to modify this code to show the position of the object being moved with the transform axis?

[Kunda1]

@Chris_G would you be interested in mentoring a GSoC applicant to help migrate functionality from Curves to Surface WB ?

[Vincent B]

First extend curve doesn't works very well at a start side.

[Chris_G]

First extend curve doesn't works very well at a start side.

The curve is wrong at its end.
the last two poles are at the same position, so the 1st derivative goes down to zero, and its direction becomes undefined.

Code: Select all

e1.derivative1At(e1.LastParameter)
# >>> Vector (0.0, -4.784750643693769e-06, -2.940684252905612e-05)