If you’re keeping up with the 3D industry, you may have heard about generative design. It’s part of the evolution of the way things are made from 2D to 3D, and it’s the next big thing that’s giving designers the superpowers to make things. Bruce Blaho defines generative design as letting the computer do some of the heavy lifting in coming up with the ideas. Bruce is the Chief Technologist for HP’s Workstations global business unit where he is responsible for driving business growth via new innovations and technologies for HP’s most powerful set of personal computing systems. On today’s podcast, he joins Tom Hazzard and Tracy Hazzard as they dive into this concept and explain how it’s impacting rapid prototyping and digital manufacturing.
Watch the episode here
Listen to the podcast here
How Generative Design Has Radically Impacted Rapid Prototyping and Digital Manufacturing with Bruce Blaho
We are going to talk about generative design 3D print end-to-end or 3D design end-to-end. We have Bruce Blaho. He is an HP Fellow and the Chief Technologist for the Z by HP business unit. His focus is on innovation and the development of new growth opportunities. He joined HP in 1985. We’ve had some guys and some women too that have been newer to the HP Community, but we’ve got a veteran here who’s been there and seen a lot of different things over time. He’s going to talk about all those changes as well. He worked on the then-emerging field of 3D graphics and then he’s moved into championing new innovations, including DreamColor monitors, HP RGS remote collaboration software, Linux space workstations, and so much more. His focus areas are in data science, AI, human interfaces, edge computing, generative design, and digital manufacturing, all of that in one. That’s a big package. He’s been doing it for quite a while and clearly, he’s got quite some resume. As an expert, we’re excited to talk to him.Before we get started, we talk a lot about iterative design, which is different than generative design. Bruce is going to talk about what generative design is, but iterative is when you make a small change in your design, print it out, and get it back. It is more trial and error. That’s a difference. Let’s hear what generative design is all about.
Bruce is an HP Fellow and Chief Technologist for the Z by HP Business Unit. His focus is on innovation and the development of new growth opportunities. Bruce joined HP in 1985 to work on the then-emerging field of 3D graphics acceleration and architected the first GPU design to run inside a window system.
Most of his career has been in the Workstation business, from the early RISC-based days to current PC-based approach. He has championed many new innovations including DreamColor monitors (which won a technical Oscar award!), HP RGS remote collaboration software, Linux based workstations, and many more. His current focus areas are Data Science, AI, Human Interfaces, Edge Computing, Generative Design and Digital Manufacturing.
Bruce, thanks for joining us. I’m glad we’re talking about generative design because we’ve been talking a lot about process flows and other things, but this is a process flow within a section of the workflow. It is exciting that we’re going to be talking about this.
Generative design is exciting. It’s part of the evolution of the way that we make things from 2D to 3D. Generative design is the next big thing in making things. It gives designers superpowers.
For those of our readers that have not heard the term before, can you help eliminate the difference in what generative design means?
The idea of generative design is to finally let the computer do some of the heavy liftings in coming up with the ideas. CAD software traditionally is about doing drawings or 3D models, but the designer had to come up with all the ideas. The idea of this and why they call it generative is that the computer is generating some of the design ideas for you.
Does this fall under a form of AI then?
In some ways, it does. Over time, it will probably be more heavily use AI. Some of the methods are a little brute force. You tried methods, evaluate them, see what works and what doesn’t, then you go back and try again. It’s highly iterative these days. The hope and expectation over time is that we’ll be able to use AI in deep learning, in particular, to try to drive to a good answer more quickly.
Give us an example of how that might be used. We had talked in another episode about a wind flow over a car or something like that. What would be a good example of the generative design working for you?
The idea is that a designer can input to the system their goals and constraints and then the system will come up with multiple options for them to evaluate. For instance, you wanted to design a bike frame. There are some things that never change with a bike frame like the seat goes here and the wheels go there. You would specify that to the system. Probably one of the first products on the market was Autodesk Fusion 360. The way they do it is you specify, “Here are the things that can’t change. The seat goes here and the wheels go there. Here are the keep-out areas. The wheels have to be able to spin here, so don’t put any part of the frame through there. Here’s how strong it needs to be. I want it to be able to hold up to 300 pounds and I’m thinking about making it out of maybe aluminum or titanium.”
Once you’ve given all of those goals and constraints, then the system can go off and come up with thousands of ideas. It will evaluate them first and say which ones are the best. It will offer you a bunch of choices and then the designer gets to go through and evaluate which ones maybe are the cheapest or strongest, which one looks the coolest. The human decides from there which ones to select and take forward.
The human is important in this process. I remember when I saw an early presentation of Fusion 360 and they were showing designs of airplanes being done by it. They were doing it in real-time live at a presentation. I was giving a talk at the same location. That’s many years ago. I remember looking at that going, “People are not going to fly in that. There’s too much open area. There’s enough fear flying as it is. This isn’t going to work.” This is where you have to step in and look at the results where you understand there are benefits to it. As a designer, you have to step in and say, “Consumer acceptance, this isn’t going to work with materials that I want to add or processes that I want to do or how I’m going to make it.”
Generative design is exciting because it does help with a lot of the engineering optimization side of things such as cost, less material, and optimal designs. One of the things that we’ll see evolve over time is doing a better job of the aesthetics and getting to know your personal style as a designer. It doesn’t do that these days, but that’s the aspiration for maybe the next generation or two is for it to learn over time based on your choices. That’s where the AI aspect of it needs to develop a little bit more. We’ll learn from you and say, “Tracy likes design stuff that looks like this. Tom liked the design, so they need to go more like this.” It will then start biasing the suggestions to what it thinks you’re going to be more likely to like.
I bet the real creative designers, more industrial designers like the idea of this too much. I understand the benefits of it from an engineering perspective. I think this may be where the word design can mean different things to different people. To me, this is the little-D design applying to more engineering decisions, and less of the big-D aesthetic design at least nowadays.
We talked to our 3D print audience and that’s who’s reading. A lot of that when we looked at what’s going on, sometimes the optimization of things, we get design sent to us and people show them to us. We’re like, “It’s great that you can 3D print it, but you’ll never be able to manufacture it.” Having that bridge that helps you understand what would it look like if I could manufacture that?
Having that interim could be useful, especially for rookie designers. That’s what I think because we have many years of experience. We have done this and failed. That’s how we learn. That’s exactly what we were talking about with Bruce. That’s how the computer learned. The same thing when you’re young and you’re starting out. That could be a great benefit to be able to use the computer as a collaborator in the process and an advisor.The idea of #generativedesign is that the computer is now generating some of the design ideas for you. @hp @zbyhp Click To Tweet
I’m a big fan of backyard barbecue and Aaron Franklin Barbecue in Austin. It’s extraordinary. He always says, “You only learn to make good barbecue by making bad barbecue.” One of the advantages of generative design is that when the computer goes through and evaluates these hundreds or thousands of options, each one of those does a preliminary lightweight simulation and test to see and you tell it. One of the constraints you give is, “What manufacturing process am I going to use? Is this going to be injection molded? Is it 3D printed? Is a cast?” That will do some testing. Part of the advantage of this is it should save you time on the backend and be less likely that you’re going to design something that can’t be built and manufactured.
I could see one application from our experience that this would be viable for. We’ve designed many office chairs that are on wheels and they almost all have a base that has five wheels and five points coming out from the center for stability. Back in the 1950s and 1960s, there were four wheels on them. They weren’t as stable and they added a fifth point. A lot of these bases we designed would be injection molded out of recycled nylon material partially for costs and ease of availability.
A lot of times, we would come up with a design aesthetically and then make us sample injection-molded tool for a couple thousand dollars. It’s not very expensive. We would test it to prove out that it would pass the rigorous testing requirements that it needed. When it wouldn’t pass, we would make educated guesses with a team of people, “Where do we need to thick this rib here or that wall thickness there?”
That’s when instead of generative, it became iterative because it goes back and forth.
If you can program in those material properties and identify them and fix those caster points, which aren’t going to change, then the computer could make suggestions and say, “If you want to withstand a load of X, then here’s what you could do.” I could see that being tremendously invaluable. Is that the idea of what something the software can do first?
That’s exactly the idea. For instance, I personally spent a fair amount of time with Autodesk when HP was first getting into the 3D print business, making sure that we were connected to all their software as we were designing multiple materials, all the nylons that our 3D printers use and all the different varieties. All of those could be built into the software and that they would know the exact physical properties of those when somebody does design and it’s not, “I’m going to 3D it.” It’s like, “I’m going to 3D print it. I’m going to use this specific material.” You can then do those iterative designs on the computer to say, “If I need it to stay on this load, and I want the safety factor of this much, how thick do I need to make those legs so that they don’t break or that it’s stiff enough?” That’s exactly the idea.
I love the way that we’re starting to look at all the different pieces and parts based on doing the series with all of you at HP. We’re starting to see this collaboration between all the different pieces. I’m thinking, as your experience in bringing in data science, AI, generative design, and digital manufacturing and all of those little pieces starting to come together, we’re creating a great information loop.
That information loop is going to inform. Often, in the design process, we finish our part. It goes over to manufacturing. Lots of these changes are made, but it never comes back into our CAD system to tell us, “This is what it ended up being,” because at that point, it’s out of our hands. We’re learning, but it’s external learning. We were bringing that back into our old system. I love the idea of that collaborative, connected environment. I’m bringing that all together in the future.
We tend to call that 3D end-to-end design. That’s the dream is I can go all the way from the design, through the considerable work to prep for 3D print to doing the print, to deploying, to shipping, and tracking. They have to be able to have all of these tools connected.
I’m going to VR and AR. We’re even going further out. That’s going to create a much richer, but an experienced environment so that we can achieve success faster.
Bruce, could a designer who maybe is more of an aesthetic designer and more of an intuitive engineer, but not necessarily a degree in engineering use this type of software to create more of understructure or frame for something for which then they can skin or put more of the outer casing covering on something? Are you seeing that application being used here?
That makes sense. Part of the idea of the generative design tools and AI in general that I love is it’s moving more of the grunt work, more of the core engineering nuts and bolts into the machine. It freezes up the human to do more of the big-D design innovation, thinking more about, “What am I trying to accomplish and higher-level functions?” rather than spending all their time on the nuts and bolts and on the engineering details. Over time, as the tools get better, it should allow people to be more creative and less engineer.
We’ve been talking about these as various things. You mentioned Autodesk Fusion 360, but there are other CAD software that are starting to bring in generative design. Is this becoming more prevalent thinking?
Yes, and It is exciting. We’ve seen announcements from Siemens and PTC. Everybody is getting into the generative design act, which is exciting because as you understand the CAD market, it doesn’t turn on a dime. It’s not always the fastest moving industry. It’s essential that you get these techniques into the tools pervasively. We’ve also seen Autodesk announced the generative design for its architecture packages for the AEC market. As you see it become more pervasively available, we think we’ll see the adoption starts to take off.
Are most of these CAD-based or are they download and can I use them on your system?
A lot of the services, for instance, in Autodesk Fusion 360, the generative design is offered. It’s fully integrated into the tool, but it’s offered as a cloud service. You pay an extra fee. You pay to do design iterations. You pay to download the designs from the cloud. The reason that they’re doing that is that the algorithms these days are brute force. You use Evolutionary Designs, where they take a few intelligent guesses. They evaluate and see what works. They take the most promising designs and use that as the seed to start the next round. It takes a lot of computing. To me, that’s one of the things that are necessary. It’s great that we have the cloud, so we can do that.
This is an iterative process, so when it’s costing you or when I have to put another quarter in the slot, every time I want to iterate, it’s a disincentive to do lots of iterating. I look forward to the next generation of algorithms that should be more deep learning-based that will allow us to apply more powerful acceleration GPUs, for instance, in particular. I’m looking forward to the day in the not too distant future when say a powerful workstation with multiple GPUs could be used to run the algorithms locally. We are right on your desk so that the engineer and designer can iterate all they want. You can do this as many times as you want and not have to pay as you go each time.
How is this important to the 3D print process and to that workflow of 3D printing designs all the way from end-to-end? What would you have to say to the 3D print readers?
The 3D print design in general goes hand in hand. It’s an interesting case where the hardware has gotten ahead of the software. The 3D printers and out of manufacturing in general are capable of creating or manufacturing structures that we can’t design in a CAD package. They either can’t be designed or they’re extremely inefficient. If I want to do some mesh or cellular organic structure to save material, it can be designed in a 3D package, but it’s extremely painful and generates monster-sized files.3D printing and #generativedesign go hand in hand; they’re made for each other. @hp @zbyhp Click To Tweet
Tom manually creates all kinds of geometry because we tend to work a lot in organic form because we also have some surface material on everything that we do. That makes it even more difficult. We understand that that’s getting cumbersome and size-wise, huge.
Generative design hand-in-hand with more modern design packages, that will be becoming and more modern geometry kernels that are being developed by some companies while you natively handle these structures. These cellular structures are organic structures rather than trying to build them all out of 3D models. It’s an unnatural act. Generative design and 3D printing are made for each other.
Bruce, I know you started a long time ago in the 1980s with CAD and I have to tell you, I did too. Although I was a teenager working in an architectural firm in a Suburb in New York City. I started as a gopher wanting to get exposed to the architecture industry, but then they knew I was good with computers and they had got in the first one workstation, then 2 or 3 that was AutoCAD. They trained me to do AutoCAD drafting at the time. In AutoCAD 2.6, that was the first one that I ever worked on a DOS machine. To think about where it’s come from these days and when we were in college getting the first seat of alias on a risk workstation or something to start doing modeling designs and then eventually 3D Studio and 3D Studio Max, and see how things have changed. We’re almost in places that we wouldn’t have believed back in the 1980s.
You were talking about the software development being slow, but to be honest with you, that distance didn’t seem possible.
We’ve come a long way. I can remember the first port of what’s now Siemens NX onto our risk architectures. I had a build of Siemens NX on my workstation as we’re reporting it over to the HP-UX operating system at the time. We’ve come so far since then. To me, this is along with the evolution of some of the new 3D Kernels or probably the biggest steps that we’re going to see in a long time.
What a great time to be in design and engineering to have all these tools available and I’m impressed.
That makes me think about the little segment of digital manufacturing, which you briefly mentioned. What’s the value for that part of the process as we start to close that end-to-end loop that we’re doing? Is it going to make digital manufacturing more fluid and easier?
That’s the idea is to have that whole 3D end-to-end to start in 3D design to be able to visualize the results in VR and then be able to have native 3D tools to allow you to do the set up for digital manufacturing. What’s exciting to me about it too then is the possibility because when a new technology like 3D print comes along and digital manufacturing, the first thing you do is you replicate everything you did before only better, cheaper, faster and we’re in that phase now.
What gets me excited then is as we get to these tools in digital manufacturing will enable you to do things that we couldn’t do before things customization. If I’m going to do a million parts, they don’t all have to be the same anymore. I can uniquely customize them. The idea of doing high volume and high variability, or even to have a manufacturing line and do a lot of different parts, that you can be agile. Today’s technology is great injection molding and things like that. It’s astounding how cheaply you can crank out the same thing in a million times, but the exciting part of digital manufacturing to me is the agility and the customization.
We used to talk mass customization when I worked at Herman Miller and that was the big catchphrase. Now, we’re moving to the opportunity of personalization and actual customization in the whole process. I’m excited about that future.The exciting part of #digitalmanufacturing is the agility and the customization. @hp @zbyhp Click To Tweet
When I started, I came to HP in the 1980s to do 3D graphics then. At that time, I was blown away by the ability to be able to draw the pictures in 3D. To be able to draw through your picture was incredibly exciting. Now, they can go all the way from those pictures, print that out as a real object like that.
Bruce, I’m glad you came and gave us an overview of what’s going on in generative design and we appreciate your time.
It was my pleasure. It’s great chatting with you.
How Generative Design Has Radically Impacted Rapid Prototyping and Digital Manufacturing — Final Thoughts
I remember reading books like To Engineer Is Human and learning that for all of the recorded time thus far engineering and design, engineering is something that it is not proven until things are tested and fail. You don’t know if you’ve engineered something properly until something breaks and if you haven’t read it, it’s still a great book it holds up to these days. It talks about design from big bridges that fail all the way down to small.
The bridges that fail, balconies on buildings or little bridges internally at hotels and they learned, “Strength of materials and stress limits and all that.” Sometimes things were overbuilt early on in the history of materials and engineering. Sometimes things were underbuilt and having things fail is how you find out if you’ve done a good job or not. It informs future design, which is iterative as well.
That’s why we feel like we’ve been successful in what’s going on because when you have done 250 designs every decade, that’s about what it takes for us to do that many designs. When you do that, or you do 600 office chairs, you start to learn those things that are no ghosts and then you start to innovate and design around that. The creativity can’t happen at its highest level when you’re worried about all of those, “Is this going to be good enough for this?” You overthink the engineering part and you get out of the creative part of your brain. We’ve not been huge fans of coding and computers generating the design, but in this particular case, it has a place and then you layer that creativity. I’m happy for things that empower us to be able to be more creative and the tools that can make that happen. There’s great power in generative design.
This is one piece of a whole process and workflow especially if you’re a more creative designer, like us, and you want to do more of the free flow creative design aesthetic process, you can use generative design for the parts of the product that needed. The example of Bruce used of a bicycle is maybe a little different because bicycles, from the entire history of bicycles, which are likely 150 years of history at this point and maybe more than that, they’ve always been an aesthetic of the material that they were created by. There are honesty and integrity in those designs.
I’m not saying there’s anything wrong with that. What I am saying is that the aesthetic and structural integrity were one and the same. They’re very much together and you couldn’t separate them. You could put some skin and stuff around a bike frame, but why would you? Take a car that has a chassis and a frame that’s going to support the engine, the wheels, and then put the outer shell around it. That’s an extreme example, but you could do that with all consumer products. Your iPhone or any cell phone you have has an outer skin and then you have all these structural electronic components underneath.
There are probably examples all over the place, but I could see how this generative design would be very valuable, especially to designers and design firms that are more about, “What’s the emotional connection the product’s going to make with people? What’s the human connection? What’s the user interface going to be the tactile qualities, the color materials, and finishes?” We can spend more time on that and not have to worry about as much of, “Is it going to hold up to the requirements?” There’s a way you can plug that in and know you have it.
When we look at this 3D end-to-end, which I love the concept of it because as we’re talking about speeding up the iterative part of the process. By doing that in that end-to-end process, by keeping all of those parts informed, re-informing, we don’t make those mistakes. We also don’t have to be present, which is how we’ve managed to control the process. With our clients, we stayed in the process from beginning all the way to sales.
Sometimes for a few months after it was first being sold to make sure that the first run of manufacturing was going well. That was how we prevented the failure points of that lack of communication through the process. It would inform us about what to do next time. We created our own success program that way. We created our own 3D end-to-end, but the reality is you can’t do that. You can’t sustain that and grow a business. It was iterative because it was pre-generative design. I do think this is a huge advancement that has a place and will be very valuable. It’s going to be a time saver.
I understand the idea of it coming back down to your workstation, especially in a large company or working on private designs and other things but one of the things that were going through my head when Bruce was talking about the cloud for AI and other things is that being in the cloud is a great stage for the development of the AI. The more failures that happen in there, the more diverse design projects that happen. We have a great friend, Betsy Westhafer, who does customer advisory boards, and she has a great podcast called Really Know Your Customer. She’s always talking about how that feedback loop with your key customers and all of that matters.
I believe in that very strongly. The data is huge is that if you’re listening to your players and to what’s going on with the customers, but in this case, when you’re trying to inform an AI, you need more data, more data is better. The broadness of it is as important as getting that one-on-one beta testing workflow that normally these software companies go through. I think that that might be a short-term benefit to the cloud, even though the long-term may not be the right thing for the power and all the things that you need in the future. We’ll see what happens with the AI. That’s the exciting part.
The way I look at it, the more connected all the different parts and pieces of the process are, and the cloud is critical to that, then the more informed everyone and every different gate in the process or node in the process is going to be. More information is better.
Remember that all of the connections to the HP and the team there is 3DStartPoint.com/hp. I want to say one last thing about this. I am learning so much because sometimes we get into our bubble. We were talking on one of the episodes that you’re getting your rhino bubble of being comfortable in our software. That’s why software moves so slowly because everyone is like, “Don’t take away my feature. I know how to use this. Don’t change my interface.” They all flip out. I learned that in color materials and finishes, when we went to go try to change the color palette and big companies like Coca-Cola and IBM would flip out, “You can’t get rid of my red or my blue.” You have that all happen here that it is that process though. Sometimes we need to have our eyes opened as to what the ecosystem is looking like for the future so that we can shift and keep up with that, but also, so we can see what’s coming on in the future, and it gives me this excitement for the next generation of designers.
I’m impressed. I think this makes a lot of sense. They get fit in with 3D printing and with a lot of the things, but when you get through this whole series, we’re doing sponsored by HP and you go through it all, it makes sense how all this fits and works together. This is critical to overall success in the industry, which I’m super excited. Thanks, everyone, for reading. We’ll be back soon with the next episode.
Get Even More!
Example Part Optimized with Generative Design
Find a Trusted Digital Manufacturing Supplier
- Autodesk Generative Design
- Siemens Generative Design
- Solidworks Design Workflow and Generative Design
- PTC Generative Design
- Function Driven Generative Design A New Way To Design Part 1
- (Really) Know Your Customer Podcast with Betsy Westhafer & Tony Bodoh
- Article in Inc. Magazine by Tracy Hazzard: The Difference Between Focus Groups and Customer Advisory Boards
- To Engineer is Human by Henri Petroski
- Bruce Blaho
More About HP
Capture and Create with Z by HP, Inspiring you for your next creative breakthrough with the Z portfolio designed and built to improve the way you create. Discover the latest Z Book to help you with your latest creative project.
Experience your design with HP Multi Jet Fusion technology and solutions reinvent design and manufacturing, unlocking the full potential of 3D printing and bringing down the barriers of 3D printing adoption across industries through materials innovation. For more details about Multi Jet Fusion technology click here.