Learning from 3D: A Conversation Between Ram Kummamuru (H&M) and Holger Schmidt

Part 1: What a Decade of Digital Fashion Workflows Taught Us

The fashion industry spent a decade and millions of dollars betting on 3D. The promise was transformative: fewer samples, faster decisions, better collaboration. But at enterprise scale, did 3D actually deliver? Ram Kummamuru and Holger Schmidt were there from the beginning. Both worked at Adidas during the early days of Digital Product Creation and the introduction of 3D for Design, Development and Downstream use cases. Both saw the promise and the reality. Now, as H&M charts a different path forward, they're willing to talk about what worked, what didn't, and what this means for the current world of 3D tools.

The Adidas Days

Before the hype cycles, before the vendor promises, there was a small group of people trying to figure out how 3D could actually work in fashion.

You both worked at Adidas during the early days of 3D for Design, Development and Downstream use cases. Looking back, what do you remember most about that period? What was the promise versus the reality?

Ram: Adidas was my first fashion company. It introduced me to the world of 3D and working with fashion and footwear. At that time, Adidas was one of the most forward-leaning brands in DPC, before Nike, before New Balance started coming along. So we were exposed to a lot of new technology, but it also means we had a fair share of successes and failures. At times, more failures than successes, because we had to figure it out ourselves.

One of the key learnings I had was about cost-to-value ratio. When you're too innovative, you often forget that ratio because you're thinking tech first, not process first. That became a thumb rule for me after leaving Adidas.

Holger: I came into Adidas from a VR/AR background supporting creation of airplanes and cars. With this background one could think fashion products are small, easy to do, but then you start understanding the domain. The approach between a rigid body like a car and a soft body like a garment or shoe is completely different. Different tools, different mentalities. The speed of creation is hugely different. From 15-20 years for planes, five to seven years for cars, and suddenly you're down to weeks.

We had to test so many hypotheses, so many different technologies. I still remember at some stage we were talking with designers about 3D and they were totally enthusiastic. Then someone opened a maker lab at Adidas and suddenly you lost half your audience because people were like, "Oh, now I can do it with my hands."

That's when I learned: there's a need for physical, there's a need for speed, there's also a place for 3D. But not only 3D. The classical approach of "we build cars and do everything in 3D" doesn't work as a 1:1 copy in fashion.

What 3D Actually Solved

3D did solve real problems. The question is whether those problems were the ones that ultimately mattered most.

Holger, looking back at a decade of 3D implementation, what did 3D tools actually solve well?

Holger: In certain areas, it solved removing ambiguity, being more precise, adding more details. In the apparel space with tools from CLO and Browzwear, designers could work on a pattern level instead of an abstract sketch level. Assigning materials, having more physically accurate representations. That certainly helped. And it reduces samples. But for me overall, the biggest difference in the early part of the process is that it suddenly adds more precision, which sometimes is needed, sometimes not.

Ram: One of the defining changes coming from Adidas to H&M was that we don't have sell-in. We have our own stores. That eliminates a huge part of the process where you need to generate images and prototypes for third-party retailers. So the core purpose of what we were working on before didn't have the same purpose here. That started shifting our understanding of where we should use 3D and why.

You can't just take any fashion 3D use case and hope it'll work across different brands. There's the complexity of different product types, but also the complexity of brand strategy. What kind of brand you are, what your core business model is. People get excited and say, "Oh, I solved this 3D problem with 10 designers working 100% in 3D." And I'm like, "My friend, we have 400+ designers. It's not the same operating model."

Where 3D Fell Short

When post-COVID financial pressure hit, the 3D value story got harder to tell.

Where did 3D fall short of expectations? What problems did brands expect it to solve that it didn't?

Ram: We had false imaginations, more than false promises, that the 3D ecosystem would be this beautiful world where everything would be smooth. But the reality was that the software doesn't talk to each other - interoperability was a constant fight. When they don't talk to each other, you have multiple applications. The cost increased exponentially. For a company our size, when you look at 300, 400, 500 users, the financial burden becomes harder and harder.

And most companies connected 3D to efficiency. Efficiency is a non-measurable thing. Very qualitative, not quantitative. When financial impacts started coming post-COVID, it became evident that it was very hard to prove whether 3D actually helped you.

Here's an example: we talk about how 3D reduces samples. That's valid. But if we reduce 5% of stock orders in the stores? That's huge compared to sample savings. I'm okay having an extra sample if it reduces the amount of stock we actually produce. The numbers just didn't justify the cost. The effort of implementation, of central resources to train people, of change management, of new tools. The overall cost, relative to the value it was generating, was overhyped.

Holger: It really depends on where you look in your organization. If you don't have (customer and internal) sell-in, is there a business case to save samples? Probably not? Then you have to make savings in the design process, in early development stages. And that's very hard to quantify.

But I think specifically in bigger organizations, there's another value people don't see: it's a catalyst. It's a change of process, of organization. We figured out the data quality (not 3D data, the actual product data) was degraded throughout the process. Thus some of the advantages we introduced through 3D and early product images were undermined by a lack of the right data at the right time. That's when we changed from "digital creation" to "digital creation end-to-end." It became a complete transformation and digitization program that was set to change how the company worked.

The Technical Designer Problem

3D tools were supposed to empower designers, but for many, it turned creatives into technicians.

Ram, you've talked about how 3D tools inadvertently turned designers into "technical designers." What do you mean by that?

Ram: I don't think designers shouldn't be technical. Otherwise it's hard for them to understand what to produce. But technical nature should not be defined because of the tool. It should be defined based on the business need.

3D became an uncontrollable habit where you cannot design without technical knowledge. That's a problem. We designed a process based on the tool, not based on the purpose of why we wanted to use the tool.

We started enforcing behavior where you have to be technical, and we stopped challenging ourselves to be sketch-driven, conceptual-driven. The tools started dictating the process, which is never how a workflow should be.

Holger: It's a question about the way of working. Some users work technically from the get-go. They have a pattern-making background, so using CLO is easy. For others on the creative side, it's quite a jump. Maybe you want to put someone next to them who has the skills. It's about the mix and figuring out where you invest.

We've recently worked with brands on automating Illustrator-based processes because it just works best for them. Technical designers moving from pure drag-and-drop to an Excel-based approach for colorization. Very technical, very non-design-friendly, you'd assume. But it cuts their process from hours to minutes, and then people adopt.

Solve the problem for the people. Don't just hand them a tool and tell them it will solve everything.

Rethinking Creative Workflows

What if the answer wasn't better 3D tools, but going back to how designers actually think?

Ram, how did you get from "3D isn't working for creativity" to "let's go back to old-school methods with AI"?

Ram: With the evolution of AI on the horizon, it started to question the base norms in how we were working. One of the major shifts we made in 2025 was questioning the tools. We weren't questioning the tools themselves. We were asking, "What kind of process do we need, and do the tools fit that process?"

We went back to basics. What if we looked at the original design world? How did design start, and why didn't it scale? You made an original sketch, but the supplier couldn't understand it. So you had iteration rounds of sampling. We asked: what if we can express ourselves better with AI?

Rather than thinking about fixing the whole 3D versus AI transformation, we took a step back and evaluated the ecosystem. For fitting, we can't use AI. We have to use 3D. For print placement, graphic placements on a specific pattern? You can't tell AI to move it 20 centimeters and scale it by 15%. 3D workflows are extraordinarily good there.

So we started looking at where we needed technical detailing for production and where we didn't. We said ideation is free-flow design. People can curate in a very free-flow manner. Development is where we've finalized what we want and start producing. Once we split those two, we were able to split the roles, split the tools, and set expectations.

Holger, when Ram first explained this approach, what was your reaction?

Holger: In simple terms: it blew me away. He cracked one of the biggest challenges we tried to crack for years. Elevating conceptual design with realistic imagery - when needed / wanted.

The speed of getting more precise, better quality by pretty much any AI model over the last two years has been mind-blowing. This was the one thing we tried to solve for years, and 3D made some dents. It has advantages in a lot of areas, but not all. And specifically not in the initial design phases when you're going creative, going left and right, trying new concepts.

I still remember walking into Adidas and seeing those big whiteboards, stickers of textiles, pictures, all this ideation and imagination happening. You could feel the creative process. With generative AI, you get to the point where you can just throw things together in a similar manner. You don't have to click buttons or draw lines. You naturally say, "Hey, I want this," and it appears.

There’s a nice example of this I see every day in my family. My daughter is 10 years old. One day she asked me to allow her to use ChatGPT to create practice tests for school. She did not type. She just recorded her ask via voice because the language models are so good now so she doesn’t need to type anymore. The barrier from clicking buttons, drawing lines, being super precise goes over to having a conversation. That's a big, big change.

Different Tools for Different Roles

AI doesn't replace expertise. It amplifies it. But only if you know what you're doing in the first place.

At H&M, pattern makers still use CLO but product designers don't. How do you think about which tools go to which roles?

Ram: We're still doing 3D for pattern making and print designs. It's irreplaceable for how to work in that space. It's not about 3D versus AI. It's about where we get the most value within the process.

Here's a fun experiment we did: we gave our AI generation tool to engineers and asked them to do fashion design. I can guarantee you, because we did it, they were not successful in making great designs. The understanding of fashion, of taste, of fit will still exist. You cannot hook a tool to expertise. They're two very different things.

Tools will keep evolving, and individuals need to evolve with them. Before we were writing on paper, then computers came, then something else. But the outcome is that pattern makers are needed. I have no idea what grading (scaling patterns to different sizes) actually involves. I work with AI day in and day out, but if someone asks me to do grading, I have zero knowledge. Only someone with pattern-making expertise can do it.

We had one designer making amazing designs with AI. Beautiful vintage retro style with specific camera angles. Other designers asked, "Why is my AI not working like that?" We realized this person had a photography background. He knew camera lenses, perspective angles. When he talked to AI, he was telling it those details.

Here's the thing: imagine trying to do that with 3D before. You'd need to set up camera angles, renderings, background scenes. Now this person can express his craftsmanship much easier. The focus should be on improvement of craftsmanship, not improvement of tools. Tools should support you in expressing your craftsmanship in the best way possible.

Holger: Craftsmanship and expertise will never go away. At some point AI will be even better, there will be better models, there will be agents and probably autonomous robots. But there's still the expertise needed to evaluate if something works, if you're making the right decisions. And add creativity, imagination, curiosity and craftsmanship.

AI is an amazing tool. It will help people go faster, better. I'm not a very good software developer. I can code faster now. Our senior developers use AI to validate their code and go faster. But will the expertise go away? No. Your job doesn't go away. It probably just transforms. You need to be able to use the tools to make your job different.

Ram Kummamuru will be presenting at The Fashion Tech Show Europe in London, March 30-31, 2026. Holger Schmidt will also be at the show and available for consultation. For more information about H&M's AI transformation or to schedule a meeting at the event, visit 4dpipeline.com or connect with Ram on LinkedIn.