Clearly this is way, way off, especially because I’m realizing this really needs to be made of hundreds of individual cube elements, all of which are “twisted”.
What’s the ideal way to replicate this Sketch Up process using Rhino tools?
I would create the overall curved shape, then come back with section planes and split up the form. It is much easier to let the NURBS forms create the smoothing for you then try and calculate it. There are many other ways also, but I might try that.
Thank you Scott! I definitely am interested eventually in taking on a more parametric/mathematic approach to this one day, but I feel like I should get more of an intimate handle on handcrafting these forms from scratch, because right now the Rhino workflow isn’t quite “clicking” for me in anything outside of basic geometric boolean style operations…
For your other suggestion, just some further Qs:
When you say create the curved shape first, do you recommend I literally only work with curves/beziers at first, twist them into a good shape, and then create surfaces from those curves? And then I split up the surfaces into small shapes (to create those hundreds of panels?)
There are a bunch of ways to do it. But, yes, creating the larger smooth shape then splitting it up later is a very common strategy. This is especially true considering the smooth transition of form is key to this design.
As you get more familiar with the process, you will be able to take any number of approaches to this shape.
The actual logic/geometry is pretty simple: First array/rotate a rectangle, then array the start and end rectangles, then split each rectangle into a glass and wall/roof polyline, then extrude these to solids:
I recently implemented this logic for an in-house Grasshopper course (note: I didn’t work on the project), as the first hands-on practical exercise. And though I can’t share this Grasshopper definition, I’d agree that Kistefos would be a perfect example of a simple project to start learning Grasshopper. As one only needs a few basic components/concepts and can avoid dealing with complex data structures etc.
I’m sure one can probably model all this by hand. But especially generating a nice falloff of the twist as it goes flat at the ends would be difficult to accomplish. I used the graph mapper for this here, as well as for the falloff along the glass openings. Also, exploring all the parameters and seeing what works is really key for projects like this in my experience (i.e. as opposed to modelling everything by hard every time you want to edit a parameter value!).
