*SUPER COOL.* A direct complete solution of the equations that describe an eight vertex polygon is by now already computable in an acceptable time, but it was nowhere even remotely possible back in 1990-1991, so the inventor has chosen the best way to do it.
I think by the early 2000s with MMX and some clever packing of values in the SSE instructions it would have been much faster. In terms of multicore it is much more difficult to optimize a single simulation for many cores, but this is void when you consider you have a population of legs anyways. You could do 32 at a time with relative ease nowadays for 600 bucks worth of CPU. Spend some time making a CUDA variant and you could evaluate thousands of designs per second. I think the real problem is the initial design. How do you come up with the geometry in the first place? That is the true genius of this wonderful man.
I am so glad to have found videos of your beautiful, elegant strandbeests and of you sharing the possibilities of wind movement creations! Wonderful! Thank you!
Theo Jansen has a talent in naming things in a memorable way. But that sometimes is an issue, because when he ships the parts of a strandbeest internationally, his way of naming the box content raises question in the eyes of custom officials. (so usually the shipping company fills out the paper work with much more boring description) Hhmm... Could be fun to see an exhibit with Theo Jansens box content deceleration next to the customs box content declaration next to it.
imagination, it puts all human's tools and faculties to work, in order to achieve a complex goal, with one simple vision, theo imagined that machine first, his brain knew it was possible to translate a brilliant thought into the physical realm, and so he did... thats the lesson and inspiration i got from him/his work/his achievements. cheers to all visionnaires !
I know. If the world had seen these beasts then and saw how he did it would have blown our minds! I first saw this 8 years ago and i was fascinated. Still am. Just amazing. And all the different kinds and sizes. Its just amazing to me.
Superb! In addition to the "flatness" of the lower curve, did you also give evolutionary advantage to a construction having "constant linear speed" for a "constant angular velocity" of the crank? I can imagine that some proportions shall give a flat bottom, but a jerky speed-profile...
BTW - more recently, (2013)the Disney Research channel has published a video about generative design for linkages and gears to mimic animal movements: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-DfznnKUwywQ.html
I'm not an "Engineer", but, I'm thinking about how to build some sort of mechanical "walking/lifting" device, that could be used to move large stones, in delicate landscapes, to position boulders, as needed, without heavy hydraulic machines!... (hand operated, using steel tubing frame, with some sort of chain-fall drive mechanisms, etc., and the ability to turn corners/walk along a curved path, or change directions/reverse course, if needed. [taking small steps, carrying big/heavy rocks.]
It's funny, so many of those measurements look so close to particular proportions that it makes my mind itch a little at the fact that they're not. [GHI] looks like it should be a 3-4-5 triangle, [ABE] looks like it should be a right isosceles triangle, and line segments a,b,c,d,f, and g look like they should all be equal to eachother. But the engineer in me knows that adjusting all of those to be "ideal" would probably quit working at best and tear itself apart at worst.
Amazing! I believe Disney or Pixar now have the technique to solve the problem in the other direction. They are able to generate the necessary lengths and joints to approximate a desired input curve, but this was only a recent development. Amazing that he solved this all those years ago. Are all the strandbeest based on this leg? What about the caterpillar/worm-like ones?
I saw that video on the Disney Research RU-vid channel. They disable comments for some reason. From what I understood they essentially create a database of every possible foot path/trajectory and then use a gradient descent method to fit the user draw curve to the closest match in the data set. While that is awesome and the results are fantastic that isn't quite the same as solving the problem in the other direction. It's solving the problem in the forward direction enough times to have a database so that you can match the desired solution to the correct input.
That's awesome that you had an Atari 1040. Maybe it was the 512? I had the 1040ST back in the day. If I recall correctly, it was the first mouse driven computer that operated in color. Figuring out your Strandbeest legs was a wonderful use for that machine.
I wonder if Theo has ever considered extending the mechanism with a second dimension. It would allow for really cool hexapod locomotion. There appear to be some constraints. The central pin is stationary. That would need to be a ball joint. There would need to be a second rotational movement at a perpendicular engine. Etc. It may be worth looking at.
The model he is using does not have section "a" AND "e" from the diagram. also what is the distance between both anchor points. I want to use my 3D printer do replicate this. Thanks
Hope you got to run some kind of patent or copyrights for both, your Atari programing and your amazing 🙂 structure designs. Your have merged science and Art so beautifully.
Patents and copyright only get in the way of science/creativity (I mean hell look at disney), Im happy he gives future inventors a easy way to improve upon his work even after he is no longer here.
Did you know that a (5, 2, 1, 0) linkage type could yield a curve with a flat bottom or did you try other 8-bar topologies as well? Did your code ensure that each of the 1500 length combinations would yield a complete curve or would some of them create combinations which were physically impossible meaning that they could not produce a complete curve?
Note his demo and drawing are not quite the same. Elements ‘a’ and ‘l’ are missing from the demo. Wonder which is Theo’s preferred implementation? I’m guessing ‘a’ and ‘l’ make the motion smoother, but with more friction.
He generated a code in a atari computer? Nice. it's interesting how much we take stuff for granted nowadays. I can just download some 3d software to be able to visualize something I thought about.
@@notcentervillewalter My goodness! You're right! Remove everything I just said. That guy was a phony and should be stripped of all his merits and contributions.