From Concept to Creation: The Real-Life Integration of AI in Soft Goods Design

August 30, 2023 Admin Admin

Embracing AI's Creative Potential in Soft Goods Design

In the ever-expanding realm of Artificial Intelligence (AI) technology, its presence and availability have become impossible to ignore. Professionals in various industries, from grammar coaches to image creation, banking to music composition, have embraced AI as an integral part of their work. AI has firmly established itself and shows no signs of disappearing anytime soon. However, amidst the excitement and potential of this new era, there are also concerns and uneasiness about the future it may bring.

One question that arises is: How can these tools be the most useful? A Stradivarius violin is just an object, but in the hands of an expert can evoke powerful emotions. Likewise, AI is a tool that can be powerful in the hands of an expert, and chaotic and confusing in the hands of a novice.

Within the creative fields, there is increasing fear and apprehension with each progressive release of more generative AI tools, as much of the perceived value of creative thinking is automated by these new tools. As they become more prevalent, it becomes critical for creative professionals to explore when and how to effectively utilize them.

As an experiment, our Soft Goods Design team decided to explore how generative AI tools can be utilized within our current design workflow. By examining their strengths, limitations, and potential for growth, we wanted to gain a comprehensive understanding of the application of AI in this specific field.

Step 1: Brainstorming and Concept Generation

The significance of rapid ideation achieved through AI technology is arguably unparalleled. The ability to generate concepts swiftly and effortlessly is truly remarkable, with effective prompt writing being the sole limiting factor.

How We Used It:

Our process began by utilizing our Stable Diffusion platform for broad image generation. This platform allows us to cast a wide net and quickly identify the key words that influence the generated images.

We started with simplicity and conciseness in mind. By placing the generated items within a background, we can provide a contextualized representation in a “real world” setting.

With the ability to generate hundreds of images within a short period, it is important to plan and group results accordingly. This is where the organization, storage, and preparation of files become crucial. This enables us to view emerging trends based on prompt type, seed image, and the number of iterations performed.


  • Quick ideation.
  • Colorway exploration.
  • Make micro adjustments quickly.
  • ANYONE can use it to generate ideas.
  • Ability to showcase “cross-industry” design elements and functions.


Where Soft Goods Design Expertise is Necessary:

  • Prompt generation. Provides a starting point and direction, guides the AI toward producing desired results, and makes the process more efficient and focused.
  • Concept viability. Helps assess the feasibility and practicality of the generated concepts.
  • Privacy and data security. These concerns are evolving and are taking crucial “fair use” and privacy questions into the court system as we speak.
  • Cross-industry knowledge. Some of the best/most interesting concepts were born from “cross-industry” prompts.  A deep understanding of prompt writing and strong image descriptors is needed.  Beyond that, knowing which industries to include can yield incredible results. For instance, using Tesla as a design reference for tents yielded angular, balanced, motional concepts.

Step 2: Concept Refinement

Most ideas brought forth in a brainstorm are discarded. The few that do make it past initial scrutiny usually require many rounds of refinement to become an actual product. By utilizing the new tools integrated into the Adobe Suite, our soft goods designers made informed, experienced-based edits to concepts at high fidelity levels.

Throughout the refinement process, we often bounced concepts back and forth between an AI platform, generative Photoshop, and hand drawing adjustments. While initial AI output was significant, every concept needed hands-on refinement to be viable.

Once we arrived at a family of key images, we brought those into Adobe Photoshop and began the process of refinement using its new generative tools. Whether we were spot-fixing, creating new elements, or completing an image that may have been cropped, we were able to take partial advantage of the functionality of these tools.

We’ve found, however, that we often need to manually fix certain things such as represented material, perspective, and structural integrity, as the AI generative content sometimes doesn’t follow what can be created in the real world. As we became more familiar with the tools, the refinement step often became a back-and-forth process between Photoshop, manual fixing, and Stable Diffusion. We figured out what aspects we liked about each program but found that there was not one program that successfully filled all our needs. With these tools still being in their nascent phase, it’s not necessarily faster.


  • Quickly make endless variations of a base design.
  • Match current rendering aesthetics.
  • Swap out materials, textures, and colors.
  • Placing concepts in a setting to help contextualize the product.

Where Soft Goods Design Expertise is Necessary:

  • Assembly and construction details. Something may look right, but when you start digging in, unrealistic characteristics arise. Surfaces that seem realistic reveal themselves to be impossible structures.
  • Hardware expertise. Knowing what’s commercially available expedites development and aids in material sourcing.
  • Material placement. A deep understanding of how materials perform over time and in certain circumstances is imperative in developing a product that will last and work in the desired way.
  • Size and scale. Real products must live and function in the real world. AI-generated images do not always account for realistic proportions and overall footprint.
  • Feasibility. Knowing when a design is not possible to make.
  • Interpretation. As you can see in the images above, the form factors generated by Stable Diffusion were interesting and visually appealing. However, many of the concepts needed an experienced soft goods designer to translate these into realistic products due to strange seam lines, non-sensical hardware and strap configurations, and Mobius strip-type construction.

Step 3: Client and Manufacturing Communication

This is where AI falls drastically short. Having accurate renderings, drawings, and prototypes is key to a seamless manufacturing handoff. If your patterns, Bill of Materials, and tech pack perfectly match, you are starting off on the right path. While AI was helpful in the initial steps, much of this crucial phase of product development requires experience, clear communication, and deep knowledge of materials, construction methods, and manufacturing constraints.

The handoff to manufacturing is almost entirely manual and experience-driven, and there is not an AI solution for this… yet. Using AI to interpret a workable pattern set from a 2D AI-generated image is impossible at this point. We begin by drawing a rough pattern directly on our image and can quickly understand where the seam lines will become general shapes. From here we go straight into our Accumark patterning software where we create the pieces that will eventually become the product. By creating our pattern piece in this software, we can quickly and effectively create patterns that will work, or fit well together.

Points of Interest

Privacy and Data Security. Asking any web-based AI generator to create client work can open you and your firm up to a host of privacy and data security issues. For instance, Mid Journey allows all users to see all queries on the platform. Are you or your clients comfortable letting everyone view the concepts you are working on? Many of our clients require us to follow a strict data security policy, so our team built an internal platform leveraging Stable Diffusion to keep any work within a rigorous data security policy.

Prompt Engineering. This has been one of the key takeaways from this activity. Often, the generator can take a lot of coaxing and strong-arming to get satisfactory results. As a response, universities and online education platforms have started offering classes centered entirely around prompt engineering. As the saying goes, “The tool is only as good as the hands that wield it.” If you want to be an AI power user, you will have to learn a new language. We expect that to be acquired through classwork as well as dedicated practice.

Intellectual Property. IP is one of the most complex topics surrounding AI. If AI generators are being trained on real-life products, do you run the risk of unknowingly infringing on an individual’s IP? Is this any different than pulling a mood board with existing products for a traditional brainstorm? Who is liable if an AI platform generates a direct copy of an existing product? We are at the beginning of these discussions.


It is undeniable AI will forever change nearly all professions in one way or another, with creative fields being one of the most vulnerable. While the current tools are frighteningly powerful, there is still a long way to go before they can subvert human guidance, creative inspiration, and intentional security.

As designers, we must maintain our traditional tool chest of skills, while simultaneously layering on new skills needed to effectively harness this emerging technology.

Laura Lenhart

In my role, I work to create thoughtful, compelling, and realistic products both in the sewn and traditional hard goods product space. With three years of experience in design, development, and prototyping, I have gained valuable insight into the gaps between conceptual designs and product reality. Understanding current available materials and processes continues to strengthen my abilities to translate ideas into realities. I am constantly interested in how we navigate this time of blurred lines between the physical and digital worlds we inhabit.

Laura Lenhart

Soft Goods Designer / Product Designer

Daniel McKewen

As a Sr. Soft Goods Designer with many years of experience I have my hand across many portions of the design and development process. Combining digital drawing, manual and digital patterning, prototyping, and vendor communication are a part of my daily routine. Understanding and implementing new processes and technologies is key to staying on the forefront of this every changing field. Priority Designs does an incredible job at fostering that development.

Daniel McKewen

Sr. Soft Goods Designer

Want to receive new articles and updates? Join our email list.