SLICE AND DICE

This is my fourth post, and represents the fourth iteration of a design process.  Stage one was just an exploration of scale, done the Revit way.  Second came the first design concept, "pumpkin in a square".  Learned a bit about using a parametric family as a design tool, but ultimately a failure.  Stage 3 was the octagon with spokes.  Much better, but is it good enough ?

What we need is unity, integration, pumpkinicity through & through, plus middle eastern to the core, not to mention fundamentally monumental.

I'm going to persevere with a smooth egg shape for a while longer.  It has a certain purity, and I don't want to get too distracted playing with pumpkin segmentation and bulge factors until I am confident about the supporting/surrounding elements.

A flash of insight early one morning while still in bed.  A pumpkin segment disembodied and laid down flat is a crescent.  How about arraying the dismembered parts of a pumpkin around in a circle?

I'm still steering clear of Point World.  Let's not get sucked into too much technical stuff.  Focus on design issues.

First of all, the crescent.  It's just a revolve with a start angle and an end angle.  Make it with splines and it will scale up with a single width parameter.  Each spline has 4 nodes.  One at each end and 2 in between to define the curve.  We can open it up and play with these to vary the curvature. At first I wamted to make it fully parametric, but decided not to get bogged down.  Easy enough to make manual adjustments and it reminds me to think like a designer.

This is all vanilla remember, so now repeaters, just a raidal array.    There's something odd about a circular array like this when viewed from ground level.

I started out with 5 segments appeared to be more than enough when viewed from above in parallel projection. But when seen from ground level in perspective the reverse is true.  After a bit of trial and error, I settled on 9

The next issue is about angles and keeping the egg suspended high enough.  Initially the crossing over of the end points was accidental as I played with the width of a single segment and the diameter of the whole array.

But realising that we needed to create a wide zone of structural support but eager to keep the full crescent shape visible, I decided that a crossover could work to my advantage.  In it's earliest incarnations, this idea involved placing an extra curved extrusion "behind" the points.  A clumsy solution.

A bit more trial and error led to the next breakthrough.  Take a segment that starts flat on the ground and cut away the middle with a wedge.

If we get the distances right, the overlapping ends will create satisfactory forms both front and back.  I am left with V shaped cuts that make natural entrance points and an interior space big enough to house lifts and perhaps escalators.

At this stage the family is not fully parametric.  I can't type in different numbers of segments and expect them to all adjust themselves convincingly.  But I can scale the whole thing up, so I did just that.

We have made some slight concessions to the law of gravity, but the pumpkin slices are not going to hold up the egg on their own.  How about some extra support in the middle ?  I imagined this as a thick circular wall punctuated by arches.  In family editor terms this means an extrusion cut by an array of void sweeps.

Later this developed into a double ring, and the void sweeps became blends so that we could have narrower arches cutting the smaller inner ring.  This was around the time when I completed my "Quarter Past Eight" post ... the time when 2015 R2 was bestowed upon us ... and I was able to take fully advantage of nudging in perspective view.

Let's treat this a bit like a real project.  Time is short and we need to get some images across to the client.  So let's take a break from family editor and all that parametric stuff.  For the exterior view, all we need is people and trees plus a bit of creative layering using image editing software.

The interior view also responds well to layering, and I've hinted at escalators with some 2d drafting.  A bit crude perhaps but it will do for now.

The section view is where drafting really comes into its own.  Masking regions give the impression of floor levels within the segments.  These will be exhibition galleries, a whole series of themed spaces looking over each other via triple volume edge spaces.  Levels linked by escalators and at least one of them continuing right around in a full circle.  The egg itself is an immense event space.  Lasers conjour up holographic representations of past civilisations: ancient sumer, egypt, catal huyuk and jerico.  This is the signature show, playing several times a week, but there are also seasonal events, concerts, plays, circuses, opera.

Underground there are more galleries, and extensive research facilities, specialising in the ancient middle east and the genetics of domesticated plants and animals.

Those images were created on 19 sept and that's the point at which I decided I had something worth sharing, and wrote the following.

So what is the verdict on pumpkin-land version 3 ?  Could be a goer.  With the time available I have to run with it.  I think it is strong enough.  Remains to be seen what I can make of it.  What happens if we introduce seams ... ie convert the egg into a pumpkin.  Will it remain a strong concept ?  I think it has a much better chance than the previous 2, but the only way to find out is to do it.

Actually it might help to unify the base and superstructure.  We have seams where the crescent/segments cross over, so why not echo these in the dome?

I am also thinking about materials and construction now.  The bowl of the auditorium has to be concrete. What about the dome?  I am thinking steel truss ribs supporting a lightweight shell.  In Dubai, domes are commonly made using GRP sandwich construction with a lighweight foam core.  For the moment let's just imagine an array of lightweight shells, perhaps a metre thick with a really stupendous U value.

There is going to be a seam between bowl and dome.  Let's not try to hide it.  Think of a jack-o-lantern.  You slice off the top to scoop everything out, then put it back on again.  It has a seam, express it.  In this case it will be much lower down, no matter, same principle.

Other organic references spring to mind.  Peeling back a banana skin perhaps, or the outer leaves of some exotic vegetable. What about an acorn ?  That has an upper dome and a lower bowl.  There must be a clue here somewhere to handling the transition.

Fenestration will be another big issue.  How to treat the inclined crescent shaped surfaces so that they can let some daylight into the interior ?  I think perhaps a continuous horizontal texture will be best.  Think louvres.  Or maybe a musharabiya pattern of some kind.

SECOND WIND

This work was done in August, four months after my initial efforts ran aground and conscious that I needed to either give it up or find a new angle.  One weekend I happened upon the idea of an octagon.  Suitably islamic, could match up with an 8 lobed pumpkin dome, let's see where this leads.

I'm going to jump straight into point world this time.  But of course we can nest vanilla families into Point World, so it's going to be mix and match.  Let's start with an egg shape to represent the pumpkin/auditorium and an octagonal extrusion to act as the rig for a table top hosting gardens above and below plus exhibition galleries suspended around the perimeter.

The rig will be hosted off a Vanilla GM family.  Just 8 model lines and a single parameter which pushes each line out from the centre.  Actually we will be using and "input * scale" system to make the profile scalable.

The egg is also a Vanilla object.  Three parameters this time, controlling two partial ellipses that help to defint a revolve.  Simple formulae will give me a "width factor" and an "eccentricity" to vary the shape, and a single parameter to determine size.  I chose height, but it probably should have been width (to make the linking of octagon to egg more logical).

Not to worry though because I am not emphasizing total parametric control at present. We can do some manual tweaking, in fact it will be essential.  Get in there and massage things around based on an intuitive sense of proportions.  Don't let technicalities dominate.  Think design.

I decide to place conical towers at the 8 corners.  These are GMA families with a single adpative point.  Size controlled by "Height", shape by "Slenderness" and "Taper". Stick them on the corners.  Add a sweep around the top edge of the octagon. A simple rectangle will do for starters.  Now we can "load into project" and get a feel for the relative size of the elements.

I'm not going to go for columns this time.  Perhaps arches will be more appropriate to the regional context.  I start with an elaborate contraption derived from the Neimeyer arch I created for RTC Chicago.  I thought this might be a nice blend of islamic and modern.

But what about the interior space.  How do we give this some character ?  And what about the link from the galleries to the auditorium ?  I don't think a full scale table top is going to cut it.

And how about the galleries themselves?  I decided to use a Vanilla Generic Model family as the profile for this sweep.  Something about drafting in vanilla seems a little simpler, more direct.  Maybe it's just an illusion.  By saving out alternative versions with different names I can swap between them.

Formally I want to keep this very simple, but functionally we need some well shaded daylight: indirect lighting for the galleries.  We used to say that "form follows function" but it's more like a tennis match: swatting the ball back and forth until something starts to make sense.

The window openings allow me to maintain the profile as a continuous loop while using it to define internal spaces.  Starting to think in terms of external terraces also, and spaces that overlook other spaces.

And the spaces below the galleries.  There is potential there.

I came up with a form to link the galleries to the dome.  Three levels of corridors giving access at performance times, and stout legs providing support to the great bowl of the ampitheatre.  I also developed the idea of a corridor/buttress linking each of the corner turrets to the auditorium.  This will have a dual function, structural support and circulation.  It will also help to shade the internal courtyards.  I'm not going to have a continuous table top this time.  Need to get some light into the interior.  Maybe will have a lattice-like shade structure

But zooming in on the facade, I get the feeling that the arch form isn't working.  Somehow the transition from arch to galleries above rings false.  This arch was not meant to carry anything above.  That's how Niemeyer used it, holding up nothing but the sky.

So I ditched the arch.

What about a horseshoe arch.  Keep it simple and abstract, grand in scale and it might work. .

The family itself is vanilla.  Lots of equalisation with all the parameters linked to a global scale factor so it can be resized along with the master Point World family.

It's always important to view your work from a variety of angles.  That's a classic BIM thing.  Jump between plan, section & 3d as you work.

The rows of arches are repeaters.  This means nesting the vanilla GM inside an adaptive family which gives me the ability to easily rotate the orientation.  In the end I came up with a rotation parameter because their always seemed to be one segment of the octagon that behaved differenly from the rest.

Each of the 3 rows of arches needed a slightly different combination of "number of elements" and "end/beginning indent".  There is clear advantage to be had by linking these back to parameters in the master family.  Keep all 8 segments in sync.

I'm starting to enjoy the "forest of columns" feel around the perimeter.  I'm adding a few "flat people" to better judge the scale.  Is it monumental enough ?

The linking corridor spokes add definition to the interior space.  Time to put in a little image-editor time.  If we are going to use BIM for concept design, it's important to integrate image-making into the process.  This might involve some 2d drafting within Revit, or it might involve exporting different versions of the same scene (rendered & shaded views perhaps) and combining these with a filter or two in the image editing software of your choice.

So, what have we got?  That was a productive weekend's effort and a much better result than the original "squared circle" concept.  But it's still not quite right if I am to be brutally honest.  It remains a collection of different ideas that don't quite hang together.  There are some nice touches and the internal spaces are quite compelling, but is the octagon sufficiently pumpkin-like ?  Don't we have too many straight lines ?

A work of art needs to hang together as a whole.  The central concept needs to be sufficiently strong to weld the entirety together.  Let's say we've learnt a lot, but it's time to start over once more.

ARE YOU DOMESTICATED ?

This post records my first fumbling attempts to design a monumental building that is recognisably both a pumpkin and a homage to Etienne Boullee. I started with a circle in a square, somewhat reminiscent of Wright's Prairie Planters (although this work was done in late March, long before I visited Chicago.)  The building purports to be a "Museum of Domesticated Species and Urban Settlement" aka MUDSUS or more popularly, "The Desert Pumpkin"

The square is clearly too high, so I looked for ways to drop it while keeping the whole family parametric & scalable.  At this stage we are trying to avoid jumping into Point World, so it's a bog-standard Generic Model family.  I'm using simple formulae and equalisation constraints to tie everything back to a module.   The module is the width of the square, "W".  It's an instance parameter with a default value of 3m.  So the unscaled version would pretty much fill your living room.

The thickness of the square "table top" is controlled by parameter "T", so why not make repeated use of that value.  I'm going to need some stout legs at the corners.  These can contain lifts to take people to the top of the table (Alice in Wonderland once more, "drink me") There will be gardens up there, and also down below.  Species that like the sun and species that prefer the shade.

I see this building as some kind of museum/research institute/theme park that celebrates the domestication of plants and animals and the role of the Middle East as cradle of civilisation.  I need some kind of functional programme to guide my efforts and that seems as good as any.  Links back to my previous submissions too, all that exploration of the vegetable and animal kingdoms.

My first camera view suggested I needed to drop the table even further down and maybe run a sweep around the edge.  It's a monstrous span between the corners so I need some intermediate support.  The neoclassical answer would be an array of giant columns, so let's give that a try.  Shades of Schinkel's Altes Museum in Berlin.  Seems a reasonable precedent.  The neoclassicists were not averse to an occasional Egyptian flourish, so I decided that would be more appropriate to the region than Doric or Tuscan.

To support the sphere at it's base I used a cross.  Even closer to Wright, how wierd is that ?  I only just noticed that now.  Then I had the idea of bringing a monorail into the desert, and extension of Dubai's Metro system perhaps.  That would be a tremendous tourist experience, flying over the dunes in a high speed train and finally docking at the giant pumpkin museum in the middle of the desert.

So I have this hollow cylinder to represent a station, and a gap in the array of columns to let the trains pass through.  So far so good, and I'm starting to think about the shaded interior also. 

Maybe we need some reflecting pools at ground level.

I'm feeling pretty chuffed at this point,  The whole family still scales nicely and I'm getting some promising images.  The original family is based on a 3 metre module

Scales up to 15m or even 150m without a problem.  And it's all done in vanilla so far.

But there's one nagging doubt.  Is a sphere acceptable ?  First of all it's a bit too close to the original Cenotaph to Newton.  Secondly, this is supposed to be pumpkin carving.  So what if I bring in a parametric pumpkin ?

There's no way I'm going to be able to do this in vanilla, so I have to rebuild from scratch in Point World.  I decide to use a box rig.  Don't need so many formulae and equalisation constraints now.  Normalise Curve Parameters will handle most of that.

The pumpkin itself is adapted from last year's submission.  Despite going way off the rails, I did in fact make a highly parameteric pumpkin, which featured in the background of my images in various guises.

Actually it's 2 half-pumpkins, or "Pump-Domes". Each Pump-Dome is made of segments in a circular repeater.  You can control the angle of the segments and this is linked back to the repeater.  Basically the angle is 360/no of segments.

The segment family is based on a box rig.  The angle is calculated from the width and length using basic Trig.  Tangent of Angle = Opposite / Adjacent.  I might revisit this whole set-up later, but for the moment I'm reusing existing knowledge to test an idea for a building form.

The columns go back in as repeaters, also the trees.  Everything is wonderfully parameteric and scalable.  I work up different versions of rendered images again. 

But in the cold light of day it all seems very forced, and to be perfectly honest rather silly.  No matter how I play with the "number of segments" and the "bulge factor" the basic composition doesn't hang together, and at this point I put the whole pumpkin idea on one side and focussed on my presentations for RTC Chicago, thinking "maybe I'll come back to this, and maybe I won't"