Design Computation Human
CDFAM Speaker Series Interview with Onur Yuce Gun of New Balance
Onur Yuce Gun has been exploring computational, generative and AI driven design since the early 2000s, with his role as Director of Computational Design at New Balance as well as his consulting work for a number of companies including Samsung and Kohn Pedersen Fox Associates.
Onur has lectured at academic institutions around the world from MIT and RISD in the US through to Universidad Adolfo Ibañez and Istanbul Bilgi University.
Onur will deliver the opening keynote at CDFAM in New York City this June. The following interview provides an introduction to his work, background in architecture, philosophy on designer-computer interactions, and the attributes he values in a computational designer joining his team at New Balance.
Can you tell us a little about your role as Director of Computational Design at New Balance and some of the projects you have worked on there?
Depending on the situation, I wear various hats. My initial responsibility involved demonstrating how to integrate advanced computational design techniques into New Balance’s design and manufacturing workflows. Despite starting from scratch as a footwear designer, I persisted through numerous rounds of explaining “what is computational design?”–but don’t get me wrong, this was a two-way learning process. I started from zero as a footwear designer, so there was a steep learning curve ahead of me as well.
In the R&D process of our 3D-printing platform, TripleCell, I played a pivotal role. Starting from scratch, we were able to release two limited-production shoes in 2019. I personally designed the forefoot midsole component of the Fuelcell Echo Triple.
Fuelcell Echo Triple by New Balance, limited series model released in 2019
I cannot list the number of challenges we faced, but looking back, overcoming all those helped us develop unique skills as a group.
I worked on anything from lattice design, simulations, light-weighting, data-driven design, part optimization for printing, assembly, durability, and performance.
You can consider this a designer’s dream, or nightmare, as it is virtually impossible to imagine a design that would satisfy all those needs at once.
Fuelcell Echo Triple forefoot part by New Balance, limited series model released in 2019
If I were to touch upon my teaching efforts in the company, I contributed to the development of Kinetic Stitch, a performance-oriented embroidery technique that elevated the quality of our football boots. As part of my efforts to train the next generation of Computational Designers, I utilized the algorithms and logic involved in the process as teaching materials.
Beyond that, my personal work encompassed a broad range of computational tasks, including last modifications to apparel, automated visualization systems, and bitmap processing.
As the inaugural “computational designer” at New Balance, I held the dual role of Computational Design specialist and brand ambassador both within and outside the company. Looking ahead, I anticipate that computational design will continue to gain momentum at New Balance, which will have a significant impact on our cultural and product landscapes.
A significant number of leading computational designers in footwear and industrial design, like yourself, have a background in architecture and parametric modeling. What factors do you think inspire architects to apply computational design in other fields? And what drives them to leave architecture and explore these alternative applications?
It’s remarkable to note that architects played a pioneering role in introducing and expanding the field of computational design. The design of the Waterloo Train Station was a turning point in the history of architecture.
In the early 2000s, the number of Computational Designers could be counted on one’s fingers (and maybe toes). I became part of this group in 2006, and at that time, there were only 8-10 of us in New York City.
Since then, the definition of what constitutes a computational designer has evolved, and for those who were early adopters, architecture began to feel, maybe, too slow.
For research-driven minds, it’s challenging to cope with a sense of stagnation. They want to ask more questions, find more answers, learn and teach more, and have a greater impact on the discourse and society.
Shingle façade pattern on Diagrid structure, Kohn Pedersen Fox Associates, 2007
When you have a passion for advancing technology and creating meaningful designs, you may find that architecture’s traditional ways of operating don’t always align with your goals. It used to be that only large companies could afford to have Computational Design (CD) teams. These companies typically work with clients who can afford to pay for sophisticated designs, which has led CD to focus primarily on pushing the boundaries of form-making and construction technologies, rather than exploring deeper questions of human existence. Unfortunately, in some cases, tools for experimentation have become tools for marketing.
That being said, some designers have left the field of architecture and pursued careers in academia, tech companies, or product design instead. Surprisingly, there are often more resources available in product design than in architecture, which may seem strange.
While academic institutions worldwide produce many architects due to popular architecture programs, the profession doesn’t necessarily need many graduates. Consequently, there is a surplus of architects seeking employment, leading to increased competition for available jobs. As a result, many architects with Computational Design skills seek alternative career paths, applying their knowledge to new fields such as product design or data science. By exploring new outlets for their creativity, architects make a greater impact and create more meaningful work in adjacent industries.
A ‘joke’ I like to make is, ‘Computational design, from facades to footwear, and nothing in between’… Is there anything in between? And if not, why?’
You made the right call asking me this question! I have some chapters in my Computation PhD dissertation where I link betweenness to creative processes and computation. Later, I wrote a more approachable version of this topic called “Why In-Between is the Best Place to be at.” You can find it here.
To sum it up, the short answer is that everything comes down to being in between! If I had to describe this with typography, the image below serves the purpose incredibly well:
Humaning – Human Being – Becoming – Computing (Onur Yuce Gun, PhD Dissertation, 2016)
I’d be happy to explore your question from a few different perspectives. Firstly, everything that occurs takes place in between, as the journey itself is often more significant than the start and endpoint. Secondly, during my transition from a skyscraper specialist to a footwear designer, I became increasingly interested in teaching and producing meaning through a methodological approach. I now consider myself a value-driven design methodologist who leverages computational design thinking and tools to create meaningful products.
Synthetic Natures, Santiago, Chile, 2012, Design and construction by Onur Yuce Gun et al.
Thirdly, if I were to take your question literally, I would consider the range of scales between skyscrapers and tiny lattices, including interiors, pavilions, installations, and handheld devices. Throughout my career, I’ve been fortunate to design projects in each of these scales, in different parts of the world and various roles. Ultimately, this connects back to my first point – everything happens in between, and the journey is what truly matters.
There’s a lot of noise surrounding generative AI for 2D images, videos, and text, as well as emerging research on generative AI for 3D geometry. What are your thoughts on the current adoption of AI in object design, and how do you envision generative AI for 3D impacting the future of product design?
The current noise surrounding a particular technology or trend is often a repetition of what has occurred in the past. Previous generations may have a deeper understanding of the historical context, but I’ve noticed that similar noises have emerged around other technologies and trends in the past:
1996 – Everything will be animated
2000 – Everything will be parametric
2004 – Everything will be 3D printed
2010 – Everything will be robotic
2015 – Everything will be VR
2020 – Everything will be MetaVersed (and NFT’ed)
2021 – Everything will be AI
Trends tend to have a fleeting lifespan, with only the well-reasoned, thoughtfully executed aspects persevering over time. I anticipate that the same outcome will apply to the field of AI. Rather than simply riding the wave of hype, it’s more beneficial to navigate it to achieve enduring relevance.
Today, there’s an immediate tendency to place AI technology on a pedestal. However, this is reminiscent of the past, where parametric design tools were initially thought to be capable of creating “everything.” Unfortunately, this idea was not feasible at all.
I believe that the use of AI for 3D modeling will face several challenges, particularly in terms of model fidelity and detailing. While I can envision some feasible applications of AI diffusion models in conjunction with current implicit modeling tools, I expect that AI will initially succeed in creating only basic forms, potentially ones that are structurally and materially homogeneous.
Voxelized and 3D printed Gyroid shape, colored by distance approximation, 2017
As the technology progresses, I anticipate that it will evolve to include gradients in the designs. However, truly finished designs always require heterogeneous solutions that incorporate a “human touch.” Increasing the resolution of a voxelized AI diffusion model may be useful in addressing these types of solutions, but the application will be limited based on the scale being dealt with.
The title of your CDFAM presentation, “Design Computation Human,” hints at the core aspects of AI adoption. As these tools continue to evolve, how do you envision individuals and teams collaborating with them?
To be honest, I don’t expect a significant shift to occur in this regard. As usual, there will be opportunities for exploration, experimentation, testing, validation, and so forth. The best department to contribute to these efforts will depend on people’s skills and inclinations.
Similar to personal computers, AI will be collaborative. While computers have been anthropomorphized for some time now, the same is happening with AI. The key difference is that AI is more powerful and faster, but fundamentally, its level of “craziness” is similar to that of a personal computer from the 1970s. It’s important not to overlook this fact.
Having worked with AI tools for several years at New Balance now, we’ve found that they present both challenges and opportunities. To address this, I take a deep dive into the tools and encourage the team to experiment with them, making everything conversational and determining the most value-oriented workflows. As a result, these tools have accelerated our collaboration within the Computational Design tool by integrating new opportunities with our existing workflows. When implementing AI systems into existing ones, having a good team with a value-driven mindset can be extremely beneficial.
Adopting a new CAD or design software tool often necessitates data connectivity with existing tools in a workflow to ensure scalability, reliability, process tracking, qualification, and archiving. Do you anticipate any challenges in integrating AI design tools into existing workflows, and how might these affect adoption?
The implementation of AI technologies will require a step-by-step approach to understanding our goals and how these technologies align with them. This will involve gradual rollouts, rather than immediate full-blown operations.
Additionally, effective communication will be crucial between individuals, departments, and leaders within every organization. Both the benefits and challenges of AI integration will need to be openly discussed. Even if there isn’t significant resistance to change, compatibility issues, steep learning curves, and costs will still be major concerns. Resistance to change is a common reaction in any experienced work environment.
My current approach to implementation involves taking incremental steps, which includes four stages: 1. Discovery, 2. Testing, 3. Piloting, and 4. Evaluation before suggesting adoption.
Once we reach the evaluation stage, we aim to have a proof-of-concept in hand to make our case more persuasive. This includes demonstrating how all the new components can be integrated into existing systems.
My approach is to advocate for tools and technologies that can make a positive impact on workflows, the lives of our associates, and our products. By adopting a fully value-driven mindset and presenting the problem with the same spirit, most of the challenges become obstacles that can be overcome with the help of a larger crowd. This allows us to benefit from everybody’s expertise and experience.
As for technical problems, scaling up the infrastructure will soon become easier as all aspects of the industry grow. While computing hardware manufacturing (anything silicon related, be it CPUs, GPUs, RAMs) may need some time to catch up with the demand for data, I believe it will eventually get there. We have been actively working on scalability, process tracking, and archiving, and with the implementation of AI technologies, I expect these tasks to become even easier.
While general 3D generative AI for products may be quite some time away, training AI on a limited dataset—such as performance and aesthetic data for footwear—seems like a focused enough approach for near-future implementation. What do you perceive as the key hurdles or steps (pun intended) to successfully incorporate AI into footwear design?
For me, the question about aesthetics is something that belongs to the past, because, yes, we know it is feasible. We used SGANs to generate a full prototype using machine learning and 3D printing over three years ago, and while we were excited about it, we didn’t think it was enough to create a buzz around. Even back then, there wasn’t much to hype about. I simply referred to the SGAN-generated latent space videos as “Moodboard++,” where instead of pinning multiple images on a board, you allow those images to be blended using machine learning algorithms.
Essentially, what this does is discover recurring patterns statistically, but what’s there to be excited about statistics? What appears as magic on the surface is actually a combination of statistics, automation, mathematics, and lots of computing power.
What really excited me was taking those latent space images and using them to suggest new tectonic forms. This is a way to unfold visual exploration, computational thinking and meandering.
Cloud Tectonics, 2020, Combining oil painting, machine learning, algorithmic modeling and rendering.
The performance side of design, which encompasses elements such as fit, comfort, durability, energy return, energy storage, and weight, is a complex and challenging problem to solve. It is difficult to achieve all of these goals simultaneously, and the success of AI in addressing them will depend largely on our ability to envision and execute solutions. Personally, I find deep learning systems and unsupervised learning to be particularly intriguing and practical for this purpose.
However, there are significant challenges to using these systems, including the need to collect and clean the necessary data, as well as the time and resources required for experimentation. This type of investment may not be feasible for all, leading to the potential use of less advanced AI systems for addressing these challenges.
As you’ve built numerous computational design teams at New Balance and for other clients, what qualities or skills do you prioritize when hiring individuals or assembling a team? Are there specific skill sets, experiences, or attitudes that you find particularly valuable in team members, and what kind of team composition do you believe is essential for success?
To put it simply, I am quite selective when it comes to filling privileged positions. These opportunities are scarce, and it is crucial for companies to make the most of them by hiring the top-notch professionals. However, my approach to selecting candidates might be slightly different.
Instead of focusing solely on technical expertise, I prioritize soft skills. I value individuals who possess a deep knowledge of their field, but also recognize the limits of their knowledge. This mindset is indicative of a collaborator who works well with teams and is always eager to learn and grow. In my opinion, these are the essential skills that make for a valuable team member.
Technical skills are undoubtedly important, and I prioritize individuals who are in the top 3-5 percentile in terms of skill level. However, I am not concerned with their proficiency in a specific software. Instead, I look for a deep understanding of techniques, methods, workflows, and logic. Enduring skills are not tied to a particular platform, and while the tools may change, the fundamentals remain constant. So, for example, proficiency in Grasshopper is not as crucial to me as knowledge of geometry, rule-based design systems, and topology. You can use any tool you prefer as long as you have a strong grasp of the underlying concepts.
After assessing the candidate’s core soft skills and technical knowledge, I move on to evaluate their critical thinking abilities. Just because someone has the capability to do something does not necessarily mean they should do it. I am impressed when a candidate can explicitly articulate why they are using a specific computational method and how it will create an impact. The power of persuasion lies in the ability to convey the meaning and value behind the work being done.
At NB, we have a talented CD team with diverse skill sets, and we work collaboratively to produce outstanding results. I don’t manage the team as they are all exceptional in their own right, and instead, I work alongside them. This approach creates a harmonious and resourceful work environment that I am proud to be a part of.
As a servant leader, I am committed to supporting my team in every way possible, whether it’s technically, strategically, or financially. I remove any obstacles that may hinder their progress, and together we work towards achieving our goals. Our team members support each other equally, creating a harmonious and supportive work environment where everyone can thrive.
I have had similar experiences building CD groups at KPF and Samsung Electronics America.
Lastly, what insights or knowledge do you hope to gain from the speakers and networking opportunities at CDFAM?
I am really looking forward to this event! The lineup of speakers and contributors is impressive and includes some of the most prominent thinkers and makers in Computational Design. It’s a great opportunity to hear from critical minds in the field and gain a better understanding of the latest developments and trends. Unlike other events that focus on popularity over substance, I appreciate that CDFAM is prioritizing what is meaningful. With the pace of change in the industry being faster than ever, it’s becoming harder to keep up with all the new developments. This event will be a condensed but comprehensive resource for anyone looking to stay up to date. I’m so excited that I’ll be bringing my whole team to NYC to take advantage of this invaluable opportunity!
Register for CDFAM to gain insights into Onur’s approach to computational design, and seize the opportunity to connect with him and other industry leaders to expand your knowledge and network within the field.
Onur is currently growing his team at New Balance and is hiring for the position of Computational Design Programmer II. If you’re interested in joining his team and contributing to cutting-edge projects, be sure to check out the job description for more details.
