# Connect by... Smooth?

Is there a reason that only extend by arcs and extend by lines is possible? Only that the point count result can be unnecessary when you just want two curves to extend smoothly and keep the same count, intersect and be trimmed.

The main reason is that extending a curve by a line or an arc of a specified radius is determinate, that is to say the outcome can only have one single geometric possibility. However, extending a curve âsmoothâ can have an infinite number of variations - so how is Rhino to decide what to do? With ExtendCrv>Smooth, you tell the extension where to go by clicking on the screen, but Connect is automaticâŚ

1 Like

Hi Mitch, Jonathan - there is a predetermined extension that a curve wants to follow âsmoothâ, - the `Extend` command will show you that if you donât use âToPointâ - so this might be possible to at least try.

-Pascal

Right I see now - Iâve actually never used to point - I always thought it was something like picking a boundary. I see now that the predetermined path isnât then the only solution.

So what Iâm saying is, could the predetermined extension(s) connect.

I think youâre describing what BlendCrv doesâŚ

@John_Brock - it is the âextensionâ style on Connect and Fillet. It seems like Smooth is also a possibility.

@Jonathan_Hutchinson1 - there is a bit more to this though - the extension is specifically for arcs, currently, it would need to be changed, I suppose, to ExtendCurvesBy=Line/Arc/Smooth. Which raises the question, âŚNatural as well? Like Extend? Iâll ask.
https://mcneel.myjetbrains.com/youtrack/issue/RH-55133

-Pascal

1 Like

Yeah - connect at a point.

Youâve got it. Certainly not one to lose sleep over. It would just in some cases cut a step or two out. Maybe it could be more hassle than worth, given that the smooth extensions can get out of hand pretty quick.

Maybe thereâd be a nice happy medium where you extend two lines, smoothly, and connect at a picked point in between, like when you are extending to a point. But off the bat I can expect that would be far more complicated. So equating the extend=ToPoint solution for two curves.

RH-55133 is fixed in the latest WIP

@Jonathan_Hutchinson1 - this is âinâ but not without its problems - I am not sure it will stay.

-Pascal

1 Like