Slashdot Mirror
Crowdsourcing Speeds Evolution of 3D Printable Objects
First time accepted submitter JimmyQS writes "The Cornell Creative Machines Lab, which brought us chatbots debating God and unicorns, has developed Endlessforms.com, a site using evolutionary algorithms and crowdsourcing to design objects that can be 3D printed in materials such as silver, steel or silicone. MIT's Technology Review says 'The rules EndlessForms uses to generate objects and their variants resemble those of developmental biology — the study of how DNA instructions unfold to create an entire living organism. [The Media Lab's Mediated Matter research group director Neri Oxman notes] that this could ultimately have an impact on design similar to the impact that blogs and social media have had on journalism, opening the field to the general public.' The New Scientist has a quick video tour."
When you write a headline and you're trying to be "hip" by using "buzzwords", please try to put them in the order in which they are actually relevant to what's going on.
I had to read this one twice to figure out it wasn't about "crowdsourcing", or "evolution", but rather about the fact that "3-D Printable Objects are Evolving faster thanks to Crowdsourcing". /pedant
sig not found
I went to the website, there was a bunch of boring 3D objects that could have been made by a 3 year old... Even the Thingiverse http://www.thingiverse.com/ is way more interesting and useful than that site.
Perhaps they can print themselves a new server, now ^_^
Slashdotting, your evolutionary algorithm site == an Extinction Level Event.
It is no longer uncommon to be uncommon.
This is hilarious. Thank god it is considered illegal in writing to force a person to accept a contract denying them to sue in case of conflict or other situation where legal actions are necessary . Now we can sue them for illegal EULA's.....
In order to form an immaculate member of a flock of sheep one must, above all, be a sheep.
I think this is crap. Objects look natural only because WE select objects that look interesting of resemble something we know, so by human selection, not natural selection, "strange" objects don't get evolved any further. That's all there is to the natural look of some of the objects.
Privacy is terrorism.
Any bets on how long it will take for them all to evolve into giant penises? It's only a matter of time.
What is shown in the article looks like the basic elements from the first 3D packages on the PC 3 decades back.
Any multi-piece design today needs explicit, exact shape and size control if you are going to have function and fit required for produceable & usable assemblies today.
I don't quite understand why we need crowdsourcing to come up with 3D printable models. Can't we take the 1000's of existing 3D models currently used in CAD programs and video games and convert them into a format that can be printed?
Yeah, let's read the opinions that are shat out by the +2 moderated posters. Write lots of bland truisms that other +2 idiots will mod up. This site is a fucking disgrace, infested by the stupid cunts that bravely mod down AC posts to -1.
Crowdsourcing accelerates evolution of 3D printable objects by leveraging the synergy of cloud computing resulting in a substantial paradigm shift in cost effective design.
-=[ Who Is John Galt? ]=-
Well yes and no. While you're entirely right that if the genome were randomly based it would still perhaps end up with things that look like human things because humans select them to look like those things, but it would have a harder time getting there. The real kicker for genome based growing like this as described originally in some of Richard Dawkins books is that that variation can such that the forms it creates and the similar but different forms are restricted such that it'll generally cover the design space for most living critters but not other things. For example, without ever seeing the thing, it's likely impossible to evolve a cork screw, but you could certainly make a very charming bowl. Some forms, due to this method, are seriously restricted and other forms are generally able to be stepped towards and the directions you go are limited by generally by natural growth patters. So you can get to natural patterns much more simply than you could with truly random modifications, but it is still humans selecting to get there so we obvious pick the forms we like the best.
We also select the dogs we like the best too, but we somehow can't select for dogs which have chansaws for arms. Doing things like this, generally gets you a next step which tend to fall into more natural patterns. It's not really that it makes the objects look natural, but hampers the EndlessForms into be a discrete set of forms which transverses a lot of natural looking objects.
It is no longer uncommon to be uncommon.
We did this 15 years ago: http://www.kurtz-fernhout.com/PlantStudio/
The approach and interface has a lot of similarities.
An open source version (in Python):
https://github.com/pdfernhout/PlantStudio/
Recent musical version: :-)
https://market.android.com/details?id=com.evojazz
A 21st century issue: the irony of technologies of abundance in the hands of those still thinking in terms of scarcity.
For another alternative, check out my comment here:
"PlantStudio Evolutionary 3D Software"
http://slashdot.org/comments.pl?sid=2443828&cid=37504222
For information about software my wife and I wrote for breeding 3D plants (about fifteen years ago):
http://www.kurtz-fernhout.com/PlantStudio/
http://www.kurtz-fernhout.com/PlantStudio/userssay.htm
https://github.com/pdfernhout/PlantStudio/blob/master/README.txt
And now breeding music:
http://www.evojazz.com/
https://market.android.com/details?id=com.evojazz
For the plants, we tried to use rules similar to what nature uses for most plants. The music one is more random and could be a lot better.
So, yes, they could make this a lot better. In general, what such a tool needs is support for a parameterizable model, where the parameters can be bred, and eventually the models themselves can be bred.
But with that said, I agree with all the hype that this is a big part of the future of 3D. We got lots of positive feedback about PlantStudio. We just ran out of money to keep developing it back then, and had to work for years at places like IBM Research on unrelated stuff to repay living expenses debt we incurred while writing it and related software (an educational garden simulator) and then got distracted with various life events and other projects.
Anyway, I wish the Cornell people the best of luck as long as the system is free and open source. And if it is not open source (I don't know) they should read this: :-)
http://www.pdfernhout.net/open-letter-to-grantmakers-and-donors-on-copyright-policy.html
http://www.pdfernhout.net/on-funding-digital-public-works.html
A 21st century issue: the irony of technologies of abundance in the hands of those still thinking in terms of scarcity.
First off, it doesn't require an assumption. Darwin was right, and the underlying algorithm works perfectly fine.
The problem is that random tweaking tends to require that there be a genotype, some underlying data that you change slightly. Unless the shape itself is the genome and then tweaking it, is tweaking it. You could, in theory, just pick a random shape and modify that shape specifically in predefined ways and get to any shape possible. The thing about genomes is that they can only code for some specific albeit limited things. You could in fact, look through every genome possible and not find some specific shape. It may well be impossible, to express that phenotype with any available genotype.
Expressing a genome in the same general growth expressive patterns of development will often hamper the range of possible shapes. But, a lot of those shapes you're limited to will be largely more natural looking. Take the notion of symmetry. Humans and most all animals are very symmetric and pretty much everything from Bilatera on has this quality and it's expressed rather clearly in the growth patterns of most animals. How many shapes are there though that are largely non-symmetric. And it turns out (given some limited space) there are hugely more asymmetric shapes than symmetric ones. But, how many of these shapes find themselves in the EndlessForms program? Very few. So while you can likely find twenty ways to get to a Tie Fighter or Butterfly, there's going to be about 0 ways to evolve a corkscrew.
But, yes, you'd have human selectors end up making a truly extensible forms look like a bunch of stuff humans know, but it wouldn't be hugely restricted to the natural looking stuff. Because frankly you'll have more things possible that look like a face in carbonite than a comb.
It is no longer uncommon to be uncommon.
Offer users the ability to print their favorite hotel/castle/maze for permanent display.
Some designs are quite neat and can often get vaped by server deaths/map corruption.
This is great if watching paint dry no longer excites you...