[Bf-funboard] Bézier-based geometry

[Gustav Ekenberg]

Hi,

I’ve had this idea in the back of my head ever since I started playing
around in Blender. Maybe this feature already exists in some form, perhaps
as a feature in some other 3D modelling application or as an add-on to
Blender. In that case I would really like to know about it.

Coming from the 2D world of Inkscape and Adobe Illustrator, I’m used to
working a lot with Bézier curves, so what I am proposing is something like
Bézier-based 3D geometry I guess.

I know we can do Bézier curves in Blender already, but as far as I know
they are used mostly for stuff like movement lines, tubes and loops. The
feature I would like to see is more like an alternative to how we model
objects today. Alongside the common way of doing things, working with
straight edges and then perhaps choosing the smooth shading option, I’m
suggesting a feature that lets users build objects using Bézier curves from
the ground up.

I think there would be a huge increase in productivity working this way, at
least in some cases. As you can see in these mockups, much fewer Bézier
vertices are needed in many cases, even if you count the handles:


A standard UV sphere (right) consists of 514 vertices, while its Bézier
counterpart (left) would use only 6. (30 if you count all the handles).


These two cones would look the same rendered but the standard one to the
right uses 33 points while the Bézier one to the left uses only 5. (14 if
you count the handles). The top point needs to be marked smooth I think,
otherwise the shape will gradually transition into a sharp point with
corners.


Here are two tori. Again, they would look more or less the same rendered
but the one to the right uses 576 vertices while the Bézier one to the left
would need only 16 vertices. (64 if you count the handles).

For objects with straight edges, like a plane, a cube or an ico sphere, the
number of vertices would be the same.

The benefit of having fewer vertices to start out with is of course that it
makes it easier the change to overall shape of the object, adding more
detail as you progress.

Maybe this idea already has been considered and discarded, maybe it’s too
difficult to implement for some reason or maybe it’s computationally
expensive, but from a user standpoint I think it would be a really neat way
of modelling 3D objects. Anyway, please let me know what you think!

Best regards,
Gustav Ekenberg