During the sculpture competition I was asked what would be the hardest part of making this sculpture and I answered that it would be weaving the cables between the ring and the grid without crossing eachother. But not till the sculpture was well under way, and I began to study this problem in earnest, did its difficulty become positively alarming. In the past I’ve done weaves using a brute force method – I would place the first cable, and then the next one goes in experimentally, either above or below the previous one. At first this is easy, but as more and more cables get put in place, it becomes difficult to find the correct path through the weave.
The size of the Nebula made this method impractical. Part of the problem was that until a cable was properly loaded it would hang with a deep centenary curve, and thus it would be impossible to know if you were supposed to go above or below. The bigger problem was that the ring was 20 feet in the air, which meant that many of the crosses would be out of reach and hard to see.
Two friends of mine, Perrin Meyer in Berkeley and Dan Torop in New York agreed to help and so I sent them the spreadsheet with the Cartesian points of the ring and grid. Perrin plotted this in plan view and I began to get worried – how exactly were we going to weave the cables into place? Perrin wrote a program to try different orders and rate them by ease of installation. For example a cable which could be installed without crossing any other is easy, while one which has to get woven high in the air is difficult. But after running a Monte Carlo program with millions of calculations it appeared that only a fraction were going to be easy, and then no matter what the remaining ones extremely difficult.
Meanwhile Dan Torop tried a different approach. Dan is a photographer, a programmer, and a mathematician and has used these talents over the years to produce an amazing body of work. Turning his focus to the Nebula he imported the cables into Blender and began scripting in Python. This allowed him to swing cables out of the way in 3-d space and get deeper into the central knot of the weave.
It occurred to me that if I made a small scale model of the cables, which had all the correct crosses, it could be blown up to full size, and all the crosses would still be correct. I began to hope I could do the time-consuming work in my shop and then take the model to the center of the actual structure in Dallas. One-by-one I could tie a cable onto a string and pull it through the model. This would give us an accurate weave with the cables on a small scale, and my theory was that when we teased it out, the weave should still be correct full scale.
I made a model and began putting in strings. But after installing 10 strings I gave up because the thickness of the string and inaccuracy of the model prevented me from putting them in with any confidence, I simply couldn’t tell which side of a string I was supposed to go around. And so Dan began using his 3-d model to create a recipe so that each string would be placed by formula rather than by eye. One by one Dan would call out which string to grab, and which other things to weave around. For example this is what he called out for string number 46.
frame 117 (step 116)
ring 046 grid X18
extend NE to 1A19
crossing ring 088 grid Z18 at 0/1 (10)%
via ring 380.93
crossing ring 388 grid I17 at 4/5 (81)%
crossing ring 070 grid Y18 at 0/1 (9)%
crossing ring 271 grid M6 at 1/2 (46)%
up to ring 46
Between the programming and Dan steering me through the Weave via the recipe, it took 10 days to get all the strings in place. Thanks Dan! – without your help I’m sure that we’d be still be tangled up in a giant knot!