PUMP IT UP

This is quite ironic.  I'm doing a pumpkin piece as a kind of diversion, an  aside ... a pause to gather my breath before returning the the real business.

Can't even remember what set this off.  And it wanders all over the place.

It starts with a globe & a 9 point adaptive component.  It's a way to get a curved skin instead of the normal faceted, curtain panel look.

I want it to be scaleable, so the offset in the middle has to be related back to the width of the panel.  Which means "reporting parameter".  There is a very unhelpful message that sometimes pops up.  It almost had me fooled that I was trying to do something impossible.  Slowly the memory cells kicked in.  You need to dimension to the ref planes, not the point itself.  If only the message said that.

So what makes this different from my profile based pumpkins? The fact that I can vary the number of lobes.  That's the benefit of the repeater/divided surface approach.

And now that the reporting parameter trick is working the curvature of the lobes is consistent, from pole to equater & back again.

The underlying "globe" is lofted from profiles.  So we can make high-waisted versions, "squarish" versions etc.

Interestingly the number of lobes will reduce right down to 2, which is a bit odd.  3 is quite interesting though.

Inevitably I find myself growing mushrooms again.

At this point I just placed a 9 point on the deck & played around with the points.  It started to look like a manta ray and I thought about hanging one inside a box rig.

This didn't really work but it got me to thinking about shell roofs and Alfredo's excellent session at RTC Auckland.

Which brings me back to abstraction & simplification once more.  I realised that Alfredo had simplified some of the forms for didactic purposes.  I had been trying to create curved valleys and getting odd results.  He had wisely kept the valleys straight.  (Keep it simple - applies to everything, but especially teaching)

But how would you tackle a more accurate shell roof with curved valleys ?  My first attempts produced a surface with an off-centre seam.  Revit is looking for the nearest straight line, which in a hypar is always off at an angle somewhere.  Orthogonal sections will always give you a curve.  By the way, what a shame that we can't set a transparency for lines (over-ride graphics by element)  I would have like my red arrows to be slightly transparent.

I took a break and made some saddle shapes.  Just 3 curves hung on a cube rig.  Select them and create form.  Very simple and easy to flex.

Then it struck me that this could be a solution to my circular array if I just cut it with a void. The first version I made was symmetrical so you could only have even numbers of bays.  

Took a while to get this working in a circular repeater.  You have to link up the parameters and use a bit of trig so that the cut angle and the length of the nested family match up with the radius and number of segments in the host repeater family.  Not too hard.

Then I rebuilt the family from scratch to make a single bay version.  Simplified the construction slightly.  It's easier to make the cut as a solid at first (so you can see it) and convert it to a void once you are happy.

Nest it up in the same way and we are in business

Put in a bit of work manipulating the parameters so that all the lengths are functions of the radius.  Now you have a scalabel family with variable number of bays and controls over proportions.

Time to scatter a few around, explore the variants and set up a render.
 I'm happy with that.  A nice morning's work.

So back to pumpkins.  What about using a similar approach ?

What I came up with is a variation on the box rig.  Kind of a box with the corners cut off.  Use it to hang profiles on.  Pumpkins.

Not just surfaces, but solid pumpkins with a hollow centre

Oddly I managed to create self-intersecting geometry that didn't keel over and dime.  Kind of a ghost pumpkin.  Very weird.

It gets to me that sometimes you are trying to do something that seems perfectly achieveable and Revit just gives up.  Then you attempt something really impossible and Revit goes ahead and makes it. All part of the fun I guess.

Finally I found myself making domes.  Pumpkin domes.  Garlic Domes.  Domes with a wall thickness to them.  AND very parameteric.

So I really couldn't resist making up a building for them.

Had to be a Russian style church/palace.  Just a little fantasy.

So that was a weekend spent playing with pumpkins, more or less.  First weekend in October to be precise,

REPEATING MYSELF

So yesterday we had scrambled cows tails.  How to figure that out ?  I decided that I needed to look more closely at the way adaptive components orient themselves in repeaters.

I created a family called (enigmatically) "Arrow" ... an extrusion sitting on an adaptive point.  Then I placed this onto a couple of curved repeaters and explored all manner of variations, taking snapshots along the way.

Here's the challenge.  I want to label the snapshots.  There are going to be 12 of them.  There is no text available in conceptual massing.  I could label them afterwards of course, but it's less confusing if I do it immediately.

My solution was to create a generic model family featuring model text.  Rig up a couple of parameters to allow me to type in the text I want (type parameter) and to control the size (instance).  I added a second line with the fixed text "Always vertical" and gave this a visibility control.  I also added a parameter to control the depth of the text and set it to automatically be one fifth of the height.

See posts by Kelvin & Philip for more examples of model text in components.  So then I steadily worked my way through the available options, taking snapshots as I went along.

Do I understand exactly what's going on here ?  To be honest, not really, but at least I have a useful set of diagrams to refer to next time I am stumbling around in the dark trying to figure out why I repeater is not working like I want it to.

And for the problem at hand (scrambled cow's tail) I found a workable solution.  Might as well stick with the default option (by host reference)  If you imagine yourself inside the repeating component, the path is going vertically up from the origin.  Think of yourself as diving down the path.  You have your nose on the origin pointing downwards and the path is flying upwards behind you.  When you press repeat, multiple copies of yourself will populate the path behind you, as if chasing you down the rabbit hole.

If it's a curved path, the component will orient itself to the plane of the curve.  So I was able to compensate for this by building an angle parameter into the component.

The next challenge was to have the vertebrae diminish along the length of the tail.  That's what they do right ?  Zach gave a solution to this long ago and others have developed it since.  You need a second placement point in your family & a reporting parameter telling it how far apart the two points are.

Normally the one point is fixed and the distance increases as the repeating components get further away.  I managed to make it work tolerably well, just the odd orphaned bone from time to time.

That was basically a morning's work.  Felt all boned out after that, so in the afternoon I decided to tack something completely different.  I knew that Escher was going to be my featured artist, so I thought I would have a go at tesselations.

I started out with a triangular grid and two types of floor (red & white). Make a triangular piece of floor and copy it around.  Change the type of alternating pieces.  As you move across from left to right, take one side at a time and replace the straight line with an "S" curve.  This is the essence of Escher's well known metamorphosis theme, which he repeated many times in different ways.

Finally convert one side to a more elaborate set of curves.  Now you have a shape which could be a bird or a fish. We need a bit of subtle detail to make this clear.  That's going to get messy if we stay with the floor by sketch method.  The system floor tool was great to allow the freedom of editing the various transitional forms and edge pieces in place, but for the final design we need a component, aka "loadable family" aka whatever.

Just two extrusions with material parameters to reverse the coloration of the types.  Very simple stuff, but sometimes less is more.  I rather like the end result, a metamorphing floor.

I went on to try another one.  Once again I started out with a simplified version of the basic geometry.

Then I fleshed this out a bit.  Let it build in complexity from left to right.  I didn't take this one through to completion because I started thinking about ideas of my own.

So I gave it a whirl.  Again start with a simple geometrical idea.  Then gradually transform it into something vaguely organic.   I won't pretend this is a great success, but you gain a deeper impression of the difficulties involved when you travel the road yourself.  Once again, I hope to return to this aspect of the challenge later if I can fit it in.