Sunday, October 28, 2012


This is just a quick assemblage of words & pictures hurled at the screen.  I am out of time & need to pack up my vegetable collection in nice little boxes to send to Zach.

On wednesday night I bought & watched "Snow White & the Huntsman".  I wasn't consciously aware of this topical link when I chose the 1937 Walt Disney witch as one of my 2 main sources.  Funny how things work out.  The film was rather better than I expected & it had an odd resonance.  They used the fairy story as a sounding board, something to bounce ideas off. 

In their case it was about story-telling.  For me it's about drawing.  You can call it modelling if you like, or design (which is just french for drawing anyway)  It comes to the same thing.  Visual explorations, representations, signs & symbols.  The history of story-telling goes back to the epic of Gilgamesh.  Our record of human drawings stretches much farther back to the caves of Altamira and beyond. 

We draw to understand.  I make pumpkins to better understand my drawing tools ... What is this thing called Revit & this wider world called BIM ?  How does it work, what can it do, does it have a soul, can it touch my heart ?  (Snow White again)

You can tick off various tricks & tips, stumbled upon: new techniques for making stuff ... but really it's about the journey.  Finding a pretext for immersing yourself in a BIM world, some reference points to check yourself against as you blunder through the undergrowth. 

And what is all the fiddling & finicking with images all about.  All this tweaking of angles and puttng mirrors in to reflect secret worlds.  Again it's something we constantly put ourselves through as architects.  Searching for the killer image, fretting about fonts, lining up images, leaving enough white space on the page.  It could be a pointless neurosis, but I don't think so.  There's a certain discipline about subjecting your design to the rigours of presentation.  All the while your sub-conscious is working away.  In the old days it might have been 2 or 3 days spent doing a beautiful watercolour perspective.  It's a chance to stop, freeze the design, probe it's inner workings. 

So on Thursday I set about a bunch of grapes. 

4 point adaptive component repeating from a central spine to a revolved surface.  Interesting possibilities. 

From there I started populating my window sill with stuff from my explorations. 

Maize, bananas, grapes of course.  And some in the basket. 

Grapes provide an excuse for wine glass & bottle.  Snow White's evil stepmother-witch demands a mirror ... which fills an empty corner & offers an opportunity for further reflection. 

Shortly before pumpkin fever hit, I began work on a corinthian capital. This uses an early version of the scaleable rig that blossomed into a banana bunch & a spring onion wig-hat.  So I can place that in the background.  And why not make it a golden pumpkin memorial. 

Actually it's reminiscent of the fire of London monument, except that's Doric (freely interpreted by Sir Christopher)

Friday I got the idea I had the tools to make the ultimate pumpkin.  Actually not quite.  Ran into the "can't make form" limitations almost immediately, which kiboshed my original plan based on the tomato.  It would have been very irregular & tweakable, but totally scaleable.   A tad ambitious perhaps.

Still this set me on to another approach based on threading profiles around a circle.  A full circle gave me grief so I made do with too circular arcs.  That's what Revit always does anyway when you "make form" from a full circle.

Hit a few more snags on the way, but had it cracked by friday night.  Broke off as I was getting into the face carving.  This was finished on Saturday morning. 

It scales OK but prefers to be super-large.  So I stacked 3 monsters like some kind of totem pole & stuck them in the middle distance.

In for a penny ... dug out some studies I did a couple of months ago ... fantasy towers for Dubai ?  It's all a spin-off from last year's Doric Pumpkin work.  Helps to fill out the background.

Still trying to come closer to the density of an Arcimboldo composition, I crammed in my Victor Horta handles as some kind of wall bracket in the corners of the window opening.  Maybe for hanging turnip lanterns on ?  Stuck a picture frame on the wall, pine cone dangling ...

And so to final renders on Saturday night.  Lots of "if only"s but I'm feeling pretty good about it.  Decided to leave the brash colours ... feels kind of Disney.

The second camera angle felt necessary ... perhaps to justify the BIM approach, otherwise it could have been photoshop I suppose.  Got carried away setting up all the reflection angles.  Notice the multiple Florences reflected off the back of the mirror.  Had to get her in somewhere.

Today I've been cleaning things up and preparing files for transmission.  Happily the server came back on line around lunch time, but file sizes are somewhat massive.  As always, I dreamed of packaging everything up so neatly, but ran out of time.  Managed to arrange most of the profiles in logical groups, but everything could be so much more rigorous ... in the next life perhaps.

Then I set to arranging the vegetables in order.  Made it to 3 sheets worth before I had to start uploading.  Here's the 3rd

So that's it.  I'll post some links to downloads later in the week.  Still 2 days of Eid left, but I need to collect my Austrian visa so I can see my grandson at christmas, then set to work on my abstracts for RTC 2013 in Auckland.  Looking forward to that !



Wednesday, October 24, 2012


I am running out of time.  We have a long weekend coming up for Eid.  The server room is shutting down for 3 days.  I can come in on Saturday or Sunday to submit my entry, but between now & then there will be no posting.  Believe it or not I don't have either internet or TV in my flat, just a guitar and a laptop.

So very quickly, here are a couple more experiments.  This is based on curtain panels.  I wanted to make a maize cob.  So started out with the sweet potato, tweaked the rig so the curves of the spline are very slight, and changed the profiles to simple circles.

I used a curtain panel from my "slug it out" post.  Very simple rectangle-based lump.

Cranking up the height parameter gives some sort of sea-slug, which randomises quite effectively too.

What about other pattern types ?  I did a quick foray into the world of hexagons.  Thought about how to make a bulge and came up with a 3-surface form.

First load gave me an extreme height to width ratio.  I wasn't going for a pine cone, but I'll take it.

Played around with slightly less extreme values and a bit of randomising & that was that.


Expect a monster post on Sunday, along with some downloads.  Wish me luck.


Jack Milburn, my first grandchild, was born yesterday in Graz, Austria.  The tomato stalk was a query I received a week or so ago.

I'm going to explain a couple of techniques a bit more carefully and in the process make some more "Edible Plant Objects", modify my "Bean Pod Hand" and further refine the "Halloween Face Assembly"

So first of all the "Centre Pole" trick.  My tomatoes are based on 5 rectangular rigs configured into a radial array.  Check out the picture.  The key to keeping this relatively simple is the centre pole.  All 5 rigs share the same centre pole (a vertical reference line locked to the centre of the family)

This allows everything to scale up and down under the control of a master scale factor.  But within the rigs, I have great freedom to vary the shape of the curves that control the 5 curve lobes of my vegetable.  There are no nested "profile" families involved here.  The tomato consists of 5 curved surfaces.  Each surface is defined by 3 open-ended curves.

The tomato stalk uses a different approach, basically the one I used last year for my Doric Pumpkin.  It's a vertical stack of closed profiles.  The profiles are all instances of the same nested family.  Each profile has an offset (Height) parameter & a radius parameter.  One of them has been pushed off-centre & given a slight twist.  A random intervention.

Time to do some work.  I want to take the wonky tomato and convert it into a mushroom cap.  First I dissolved all the surfaces, then I needed to isolate each of the 5 profiles, one by one.  That's pretty easy: view cube into a top view, make a window around what you want, use the sunglasses to isolate. 

I have made my rigs wider and flatter, as befits a mushroom.  This causes the original curves to bulge out rather alarmingly. No matter, I am going to rebuild them into two simpler curves meeting at a point.  Plenty of patient point re-hosting ensues.

Having done all that 5 times over, I can select curves, (3 at a time) and "create form" them into surfaces.  Separate surfaces for top and bottom so I can use different colour materials.  It definitely looks like a mushroom so  I am a happy chappy.

And what are these mushrooms for ?  Lips of course.  Which lips ?  Quite right.  Witch lips indeed.

Pushing my head "sub-assembly" back into the main model, I can make a few more tweaks.  Reverse one of the avocado eyes to get the sideways glance.  Play around with the eyebrows, hunch up the shoulder blades.  Need some teeth though.

Teeth had better be pips.  Could be pumpkin seeds, grapefruit pips, who knows ?  This has to be a stack of closed profiles.  Tried to be too clever at first & came up with a profile that wouldn't join up smoothly.  So I simplified.  Still can't quite predict when I'm gonna hit one of these "can't make form" issues.  I guess it means I'm pushing the boundaries a bit.

Nice toothy grin, more fiddling with the eyes & brows.  Facial expressions are pretty tricky.  Our subconsciouses are so good at picking up every subtle nuance.  Conscious brain is left standing.  Squint your eyes, go make a cup of tea, ask someone else.  It's coming on though.

By the way, the teeth are fully scaleable.  All the widths & breadths & heights link back to a single scale parameter.  Could have built in more variability, but I think a simple change of size is enough.  The pic below catches me in the act of filling out the formula fields ... 2 down, 7 to go.

I need the left hand to hold the basket.  So my bean-pod fingers should curl around more to form some kind of fist.  The pods themselves are nested inside a hand assembly, so copy and rename that (right hand, left hand seem good names)  Can rename the bean pod if you like, but it's safely wrapped up inside the hand so no big deal.  Ah but the magic of the rectangular scaleable rig !!!

All I need to do is rehost a view points and nudge them about, hey presto, curly fingers. 

OK so it's not a perfect fist, but it will do for now.

I'm getting close to the point where I can put the Snow White source image aside and do my own thing.  Or perhaps pay more heed to Arcimboldo.

So what else does it need ?  I'm thinking it's a bit bare.  Archimboldo's pictures are absolutely crammed with stuff.  Not sure I can get anywhere near that, but there is time to throw a few more vegetables at the canvas & see if they stick.


Tuesday, October 23, 2012


How to make a spring onion for a wig hat  ...  AKA  ... the general instructions for making an adaptive scaleable rig.

Place 2 points, one vertically above the other & make them adaptive.  Spline by points, make reference.  Set current work plane to back/front of one of the points, place another point, give it an offset.  Repeat for the other point.  Join it all up into a rectangle, make all lines references.


Let's call the original ref line joining the 2 adaptive points the "Backbone".  Set current plane to L/R of backbone.  Dimension between the adaptive points.  Make it a reporting parameter "H".

 Parameterise the offset value of the other 2 points, let's call it "D" for depth.  And we'll set up a simple relationship between Depth & Height based on a factor "F".  This basic rig can be used for almost anything.  Place points to give the curves you want.  Actually the curves shown here are from a first failed attempt.  Had to simplify second time around

Prepare a profile for scaling.  The idea here is that the radius of each instance along the path can be easily adjusted.  Just select the profile and type in a new "input" value.  Meanwhile there is a master "scale" factor which is linked to "H" in the host family.  So move the adaptive points apart and all the profiles adjust proportionally.

The profile goes on to the curves, repeatedly.  For the spring onion I needed about 10 strung along 2 paths. 

It's a stylised sprint onion, one white section that bulges out, 2 green legs emerging with constant radius.  This is complex enough without getting carried away.

The width factor (F) serves to vary the curvaciousness.

The first repeater was based on simple circular arcs, drawn on horizontal planes.  Had to set the adaptive points to "orthogonal in family" in order to avert a bout of wonkiness.


Second iteration of divide & repeat used 2 tiers of dancing green onions.  I like these little guys.

third time around I went for free-flowing splines and eventually 3 tiers.  We need to evoke the raggety locks of a wicked witch.

Now we have a wig hat, (reference to "high heeled sneakers", the music to this post is pure class, Tommy Tucker & Booker T ) so now we have a wig hat I can bring back the face.  I am making the whole head into a sub-assembly so it's easier to cock to one side in the main model.  Going after that sneaky sideways look mentioned in a previous post.  Trying hard to name my families sensibly also.
I think the butternut squash chin is a nifty move.  Tasty too.

By the way I discovered something very handy while building my spring onions.  It's an extension to the previous revelation that you can select a spline, select a point or 2 & create form to extend the line without the trouble of re-building.  Sometimes it's the only way to get points to join up in the order you want.  WELL you can do the same with forms.  Select a form based on 3 or 4 profiles, select more profiles & create form.  The lofted form will extend.  This is magic.  You made something with 7 profiles only to realise you need another 2 to get the bulges you desire.  No need to dissolve it all and start again, re-apply material parameters etc.  Is this documented anywhere ?