Ask the Experts: AI’s Impact on 3D

448

As part of our ongoing ‘Ask the Experts’ series, we have brought together some of the Fashion industry’s leading Digital Transformation specialists to answer your most pressing questions. Today, we ask the team: in what ways do you predict AI will impact our use of 3D, and how should we best prepare? Here’s what they had to say.

AI will undoubtedly impact almost every aspect of the 3D workflow as it relates to the fashion product process. Because of this, you need to focus on defining your priorities and resist chasing after all and every exciting new AI capability that emerges…there will be too many of them!

Think about: what are your major pain points, and what solutions will most impact your company’s profitability? For example:

Is the concepting and line planning process slowing your company down, or is it the material development and costing processes?
Is your team struggling with 2+ prototypes per style or is it sample shipping and photography causing the biggest delays and costs?

Whatever the answer for your company, build a roadmap and prioritise your easy wins. And do not view AI as a quick-fix for everything at once.

You should also think about how your company needs to change in their overall processes, and start thinking about how AI and 3D might support this.

Find Joshua here.

There is no need for predictions…it’s already happening!

We should be re-thinking the role of 3D and what AI can replace. I am currently working with two GenAI start ups – one in the US and one in the EU (and there are more) – to realistically render from flat, black and white designer sketches to garment simulations that not only look better than most 3D hi-res simulations, but also take a fraction of the time and are eCommerce ready (including brand lighting aesthetics). 

We have completely reduced the need for colour and pattern filled 2D paper dolls. When backed with data from trend research and predictive analytics from bestsellers, the photorealistic GenAI simulations reduce risk in range/line planning, merchandising and design review, and edit with more confidence prior to cascading styles into the 3D workflow.

We are therefore reducing and improving some of the 3D workflow with AI. 

That being said, we should not be seeing AI as a full 3D replacement. We still need 3D to create technically accurate garments for fittings and production, but AI is simply taking on some of the load for an improved and agile workflow.

GenAI also doesn’t require the skills of a product engineer/garment technician…anyone can use it! As such, we should be encouraging as many people as possible to experiment with these tools as part of their day to day to ascertain how they might best fit within your overall operation.

Find Craig here.

I already know of several projects aiming to leverage AI to significantly streamline workflows when it comes to 3D visualisation of clothing, and I daresay we’ll see some big changes within the next few years. Funnily enough, I’m one of the very few people that actually enjoys the process of digitally sewing together garments – I find it rather meditative – but on the whole, people should embrace the opportunity to automate that step. Even rough pattern creation from sketches (or GenAI images) is in the near future!

And the more we can get AI to automate part of the workflow, the more creative freedom your teams will have. Using 3D already reduces or cuts out the toile-making process, and if I, speaking as a Fashion Designer, can create a rough pattern from a garment sketch, have the program assemble and sew it together, and use GenAI tools to create fabrics and prints according to my own ideas, I can very quickly reach a polished result. I personally can do all of these steps passably well myself, but the time investment is large. Without these tools, one has to either learn pattern making or hire a pattern maker, learn how to create digital fabrics or hire someone for it, learn to create prints or hire someone for that step, etc.

The really neat thing is that this goes for other members of the process as well: a Pattern Maker wanting to show off their skills can get a GenAI tool to suggest designs, do all the pattern making themselves, and then use a 3D software to create realistic visual representations that will convince potential clients more easily than only the flat pattern would (since only people with pattern knowledge will appreciate good pattern).

For companies at a higher level, AI also promises many benefits. It’s not about making employees redundant, but lessening their workloads in areas they don’t need – or want – to be involved in. I’m sure I’m not the only one who’s heard a Fashion Designer in a big company say that they already have too much on their plate, but are still expected to add more to it. Automation and bridges between systems are great things, and just like the spam filter in our email inbox, AI can and will make our lives much simpler. 

Find Sophie here.

AI is undeniably taking our world by storm and nobody truly knows what profound implications the rapid development of this technology will have on the way our industry operates.

I personally see the rise of AI-powered tools and platforms as an opportunity for us humans to truly reflect on what we bring to the process that cannot be automated, standardised, and replaced. As a designer whose lifelong focus has been on empowering other creatives to design better products and experiences, I see Generative AI as a tool that allows us to explore creative directions we may not have considered otherwise and/or make design decisions that are informed by information we might not otherwise have access to, all in real time.

When it comes to 3D-specific workflows, our tools and platforms are bound to change dramatically, lowering the barrier to entry to people who have great ideas (and taste, hopefully) but don’t have the skills to communicate them. The way we interact with software to bring product concepts to life is bound to change as well. Instead of clicking on menu commands, we will be able to just talk to the AI assistant who will co-create the asset with us. 

The path from idea to a high-fidelity, decisionable 3D asset is becoming shorter and shorter. Text-to-image, Text-to-video and Prompt-to-3D will soon become powerful ways to accelerate the creation of 2D and 3D assets and will require a whole new set of skills that will need to be taught on top of the foundational knowledge of garment & footwear design, but will eventually streamline the DPC process tremendously. 

Although AI-powered software will eventually streamline a large part of the creation process, Generative AI is a powerful capability that requires its own, very different type of skills. It is imperative for our educational models to be able to anticipate these constant developments so that organisations can stay competitive by upskilling and hiring new talent. 

Find Safir here.

I am super excited to see how AI will drive meaningful change within apparel digital product creation, but at the same time, I admit that I have some doubts and concerns. I’ve been in the 3D trenches for a while now and apparel is a tough nut to crack. What I am hoping for is that AI unlocks better standardisation, and forces us to revive core principles of garment engineering – let’s democratise patternmaking for a start!

In less “State of the Apparel Union” terms, our industry is a high-functioning trainwreck. AI is excellent when we have excellent learning data to feed it. Unlike literature or other spaces, apparel design and manufacturing doesn’t have a concrete common “language” or margins to follow. Once we add in all the variables of our choo-choo trainwreck of an industry, I’m left thinking “gee… aye… I don’t know…”

The good thing is, AI doesn’t have to conquer every parcel of our industry at once. If we make progress a priority and bring our colleagues along for the ride, I absolutely believe great things can happen for apparel and AI. If we let our same old habits lay out the yellow brick roadmap, we may end up stuck in the trenches outside of the OZ city limits. Let’s get our act together and make this work.

For example, as wonderful as our cloth simulation engines have become after decades of development, we’re still short of successfully simulating textiles with the accuracy needed to predict production issues. We can still break cloth simulation engines with bias-cut garments. I know, I know… It’s not the 1930s, but unless you cut in perfect 90° angles (on-grain), you have bias sew lines on your pattern pieces. Physics that do not understand (or cannot rapidly compute) the relationship between yarns will continue to fall short. Then add the reality that warp and weft do not respond to gravity the same way…you may be asking why is this relevant to an AI discussion? Because many software engineers creating physics engines were never “fed” that information. We dropped the ball helping our human intelligence succeed. Let’s do better with AI. If not, we’ll just end up with AI apparel tools that misinterpret us like Siri and Alexa do!

Find Christian here.

A big thanks to our experts for their insightful and informed answers!

Found this helpful and have more questions you’d like answered? Pop them into the comments section below and we will add them to the list!