Extrude Boundary Surface - Not extruding in direction given

[martynjhogg]

I have a curve created from a rectangle with radiused corners and I have divided it with Perpendicular Frames.

On these PFrames I am creating more rectangles, then Boundary Surfaces and finally extruding the Boundary Surfaces in the direction of the Z direction of the PFrames.

The Vector display shows me that the vectors are in the right direction, but some of the extrusions are in the wrong direction…

Is this a bug or have I missed something?
Is it something to do with the curve being a rectangle?
VECTOR_ERROR.gh (11.3 KB)

[Joseph_Oster]

@martynjhogg

Is this a bug or have I missed something?

Really? Without posting a GH file?

[martynjhogg]

Schoolboy error, my bad.

[Joseph_Oster]

Works fine if you skip the Seam component. (R7)

[martynjhogg]

Interesting!
So is this a bug with Seam, PFrame or Extrude?
The vectors look OK according to the Vector Display

[Joseph_Oster]

This also works, indicating a possible flaw in Crv CP.

VECTOR_ERROR_2025Feb14a.gh (9.2 KB)

I really have no idea.

[Joseph_Oster]

This is really weird

VECTOR_ERROR_2025Feb14b.gh (12.2 KB)

P.S. Another peculiar detail - look at the location of End Points:

[martynjhogg]

I’m trying to have an odd number of PFrames symmetrically about the X-Axis so on the right there is PFrame exactly on the X axis and on the left there is one either side of the X-Axis…

That’s why I used the Seam component.

In your last definition, if CrvCP AND Seam have their Curve input reparametrized or not then it fails.

If one is reparametrized and one isn’t then the extruded is correct but they are in the wrong place.

[Joseph_Oster]

@martynjhogg

I’m trying to have an odd number of PFrames symmetrically about the X-Axis so on the right there is PFrame exactly on the X axis and on the left there is one either side of the X-Axis…

A flaw in that plan is that you are extruding in the pFrames’ Z direction, so it’s impossible to straddle the X axis. But there is more weirdness than that going on. Using ‘World XY’ rotated 90 degrees, the extrusions are affected by the the large rectangle radius. More than 30 looks OK, less than 30 skews some extrusions.

VECTOR_ERROR_2025Feb14c.gh (13.4 KB)

[Joseph_Oster]

Where there’s a will, there’s a way. Using Domain Box instead of Extrude works with ‘World XY’, rotated or not.

VECTOR_ERROR_2025Feb14d.gh (12.0 KB)

Doesn’t explain the weirdness with extrude but gets around it.

[leopoldomonzani]

Reparametrize?

[martynjhogg]

So if Seam is Reparameterised and CrvCP isn’t (or vice versa) the extrude works but then the PFrames are in the wrong place. I want them symmetrical about the X-Axis which is why I use CrvCP to set the Seam on the X-Axis.

[leopoldomonzani]

Try this solution.
I put a scroller to move the seam manually but obviously it can be aligned automatically.
vector.gh (19.2 KB)

[inno]

I bypassed the deconstruct plane component to narrow the possible culprit (Plane cast as Vector gives its Z component anyway, and cast as Point gives its Origin)

it’s weird to see that if you extrude the rectangles directly, they are not solids but surfaces to be capped, but the direction is fine?

[Tom_P]

tried a few things:

• do the extrusion in a scripting-component - strange direction
• extrusion component - strange direction
• vector Display - expected direction
• Move - strange direction
• draw a line from plane Origin - expected direction
• draw a line from shape-Curve-Startpoint - strange direction

tried to

• bypass the curve-Frame component - no change
• bypass the vector Amplitude component - no change

→ i would call it a bug

VECTOR_ERROR_2025Feb14c_extended_tp.gh (17.4 KB)

[inno]

@Tom_P

→ i would call it a bug

and probably it’s related to some weird interaction with Boundary Surface, because if you cast the Rectangles as Surfaces it works fine?!?!

[martynjhogg]

Thank you @Joseph_Oster @inno @Tom_P and @leopoldomonzani for looking into this.

It certainly feels like buggy behaviour.

For me, simply swapping out the Boundary Surface component and casting the rectangle as a surface is the easiest fix so @inno gets the solution tick!

It’s been good to learn that a plane is passed as its Z vector or Origin if it is plugged into an input requiring those inputs, although I don’t like doing this.

Should this be reported as a bug?

[Joseph_Oster]

@martynjhogg

For me, simply swapping out the Boundary Surface component and casting the rectangle as a surface is the easiest fix so @inno gets the solution tick!

But as I said:

@Joseph_Oster

Using Domain Box instead of Extrude works with ‘World XY’

Extrusion doesn’t meet your stated goal of being symmetric on the X axis

@martynjhogg

I’m trying to have an odd number of PFrames symmetrically about the X-Axis so on the right there is PFrame exactly on the X axis and on the left there is one either side of the X-Axis…

@Joseph_Oster

A flaw in that plan is that you are extruding in the pFrames’ Z direction, so it’s impossible to straddle the X axis.

[martynjhogg]

@Joseph_Oster

Extrusion doesn’t meet your stated goal of being symmetric on the X axis

Ah, I move the extrusion afterwards, but you are right, the domain box is a neater solution.

[Tom_P]

@wim or @Gijs
even if there is a “solution” - which is more like a workaround - there is some buggyness here.
hot sure who should look at it.
thanks for assigning it to the right person.