It’s 3am. Visions of green and red shapes are swirling through your mind. Was it a bad batch of kombucha? No – it’s Autodesk Generative Design.
Autodesk Generative Design provides a surprisingly addictive mix of novelty, somewhat mysterious results, and interesting shapes. If you can justify the cost – try it!
Before I dive in, I need to comment on pricing. I wrote this piece before I knew what Autodesk’s new pricing structure would be as of October 7th, 2018. I assumed that all Fusion 360 subscribers would not only have access to Autodesk Generative Design (AGD), but would be given a handful of cloud credits with their subscription.
Unfortunately, as of October 11th, 2018, that is not the case. You start with no credits and must buy them $100 for 100 credits. Thus, you can’t try AGD without up-front investment. I think that’s a mistake on Autodesk’s part, but that’s not what this post is about. Update: In the Fusion 360 October Update, Autodesk has updated their Cloud credit allocation, stating, “Trial users get 300 CC for the duration of the trial.” and “Initial Subscribers are allocated 100 CC” with a note that, “The first purchase of any subscription, for any product, receives a one-time grant of 100 cloud credits. These cloud credits do not have an expiry expiration date.”
You should know that as a Fusion 360 early adopter I pay $300 annually and receive what was previously an Ultimate subscription and 1000 cloud credits. That bank of cloud credits is what I’ve used to learn and work in ADG. To date, I have spent 650 credits in ADG and feel like I have about a 60% expertise in the software. OK, that’s enough of that!
Learning Autodesk Generative Design is Fun
First comes the thrill of learning a new design language. Gone are the splines, the extrusions, and the shells. Hello preserve geometry, obstacles, and convergence. In constructing a physical object with generative design one must think about the design in a fundamentally different way. That kind of stretching as an engineer/designer is what it’s all about! I attended an in-person training at Autodesk’s office in Portland. It was very helpful, and if you have a seminar near you I would recommend attending. Spend their cloud credits, not yours!
Since the seminar, I’ve worked on a half dozen different generative projects, all of which are violin parts. Over the last couple months, and in part to support my generative efforts, I designed the open source Modular Fiddle.
Any part of this violin can be swapped out in about 10 minutes, which makes testing variations ridiculously easy. The goal of the Modular Fiddle project is to use 3D printing and digital design to push innovation in violin sound and design. For example — what can generative design techniques do for an acoustic violin? Let’s take a look at the two parts I designed using ADG that have actually become physical objects.
First Project – a Violin Bridge
My first project was a simple violin bridge (the part that supports the strings). In total, I spent 150 cloud credits to get to the finished part.
Forces on a violin bridge are very straightforward. So my setup in AGD is, too. In AGD, geometry you want to keep (preserve geometry) is green. Geometry that represents a keepout (obstacle geometry) is red.
Strings go here, push down with this much force, and you need feet here. Let’s see what we get!
I like looking through the range of geometries, and scrolling iterations to see how the algorithm got there – it gives me ideas. ADG is a great tool for ideation, which I’ve heard Autodesk (at least informally) say is the intended use. But I want to use one for real! Let’s pony up and drop some cloud credits to bring this into Fusion 360.
In Fusion 360 I cut a flat plane on the back of the bridge for a print surface and added material to a few thin sections. I also added a strut across the lower legs of the bridge for additional support. With those changes in place – it’s ready to print.
As it turns out – this thing works! In fact, it gives the violin a very different sound. You can hear it in the video, below, in which the red violin has the generative bridge.
Level Up – Adding Complexity
At first AGD hooked me with a new design language that was fun to learn. Once I had that figured out, I experienced the joy of seeing novel solutions appear before my eyes. Then I tried some more complex generations and got poor results. I won’t go into detail, but if you’re interested you can read through a forum thread I started.
Let’s look at something more complex. This time I’ll take a stab at generating a solution for the Modular Fiddle’s Pegbox, shown above. The Pegbox is the 3D printed part at the head of the instrument that supports the tuners. In total, I spent 200 cloud credits to get to this final result.
Here’s an early setup and one of the results (I ran a total of four generations):
It’s pretty clear that this part would break. This fragile result is due primarily to an improperly applied fixed constraint. I can also see that I need to add more obstacle geometry. My fourth generation, shown below, creates some parts that look usable.
I exported a SAT file of the above design and printed it without making any modifications. It’s officially 30% lighter (weighs 19 grams) and 175% more badass than the original.
In summary – Autodesk Generative Design is fun to use, gives interesting results, and can make real stuff.
Forget the marketing talk about the future of design. Try it because it’s fun and you get to learn something new. Setting up the problems in AGD requires thinking about your design differently, which by itself has value. After setup, combing through the results is enlightening. Whether you’re looking to use the results for ideation or want to export usable solid geometry out of AGD, it’s worth your time to set up and run a few generations. Is it worth your money? That I don’t know.
If you need a project idea, let me know, I have some violin parts you can work on. 🙂
Images courtesy of OpenFab PDX.