The creative process behind AI imagery in fashion trend forecasting
We speak with Ryan Parry, Digital Asset Specialist & AI Image Analyst at WGSN, for a behind-the-scenes look into the creative process of using AI imagery in fashion trend forecasting.
Q: What do you do at WGSN?
A: Up until recently, I have worked primarily within our Image Management department, categorising imagery that provides the essential data which powers our insights across multiple areas, including Catwalks, Streetstyle, Trade Shows and Retail.
With the rise in AI, I realised the potential of what AI imagery can bring. I now create AI imagery based on image requests from our teams to visually translate our concepts.
I’m mindful of divisiveness and what AI means for the future of creators. I closely monitor and explore these emerging technologies to understand the controversies and their portrayal not only on social media but also the legalities and opportunities as they develop over time. I continue to explore these new tools and identify the risks and potential while supporting our vision of the future.
Q: Can you describe the process of creating an AI image?
A: I start by visualising how an initial concept or combinations of ideas can fit an overall narrative. For example, “a mycelium velvet shirt intertwined with metallic malachite jewels in the style of cosmic-inspired hard tailoring”. I tweak certain words that change the overall image and then align it to the overall aesthetic of the original idea by reworking the images and modifying prompts until the result fits the narrative, an emotion or aesthetic.
You can be pleasantly surprised or very disappointed by the results. A good result takes me in a direction that I can then use to build upon or go in an unexpected direction altogether that speaks to the idea while creating a new version of the original concept.
Combinations of ideas affect the results – certain word combinations may have accurate and positive results or the idea may not present the desired outcome. It’s similar to putting paint on canvas; it’s the layering of ideas that emerges a coherent image.
The process is highly individualised, based on feelings and personal taste, resulting in endless variations. Inaccuracies can occur since AI image tools, even with recent improvements, are still in their infancy.
Also, as a neurodivergent individual, it has given me the ability to articulate my own creativity in ways that I’d never have imagined before. It’s like being handed a golden ticket to Charlie’s Chocolate Factory, where there are so many treats to choose from giant jelly bears, candy apple trees to the chocolate river. The infinite possibilities enables me to express my imagination through image creation.
Q: Compared to a regular image, how does AI imagery compare? What are the benefits?
A: AI-generated images are fantastic for articulating ideas and accelerating concepts like never before. Despite its limitations, the recent developments in accuracy, quality and speed make it a valuable resource worth embracing.
As these tools evolve with more granular customisation for colour, specific details and accuracy, I anticipate they will become necessary tools across industries to facilitate concept development allowing creators to take more hands-on approaches with other skills and work on their craft.
These tools won’t replace nor discredit beautiful, original imagery but we live in an AI-focused future and must adapt to emerging tech.
Q: Why?
A: AI image companies used publicly available imagery to train their models, causing divided opinions and tension across the creative industry on creative and social media platforms. Many terms of service agreements give exclusive rights to use uploaded content as they wish in exchange for the free services they provide.
Most often ignore the fine print when sharing copyrighted material online. It’s complex and blurs the lines between personal ownership and public use. To navigate this, we must harness opportunities to develop new skills to our advantage. With the ever-increasing gamification of the attention space and algorithmic influence on trend cycles, there’s a stronger need to ‘get it right’ at speed.
Q: Is AI Imagery just a fad?
A: From what traditionally would have been the slow development of an art movement, the access and sheer volume enabled by AI can cause a paradoxical effect that can lead to regular burnouts. We understand the cognitive fatigue from the habituation of novel content and how the oversaturation of this content can seem overwhelming – especially from a social media standpoint – that can reduce our perception of value and authenticity. At WGSN, we aim to showcase the benefit of these tools with the prospect of scaling back physical prototyping and provide clarity on concepts that allow innovative approaches to product development with greater focus on strategy, efficiency and cost-effectiveness in design processes.
Q: What role does AI Imagery play in WGSN’s trend forecasting process?
A: AI imagery serves as a supportive tool for visualising our STEPIC narratives, key colours and trend ideas. Combining our wide and varied concepts from our global research, it allows us to showcase these ideas as potential realities. WGSN is the leading trend forecaster unfolding ideas across industries every day, AI imagery supports this offering with an added value that gives our clients never-seen-before visuals that we hope inspire a better successful world.
Q: What is the future of AI imagery? And how can WGSN clients use AI Imagery to plan ahead?
A: We understand the evolving ethical and legal landscape around AI imagery. We have a strict policy that allows us to use emerging AI image technology to visualise our narratives as inspirational supplements to our content. When prompting, we never reference existing creators. Instead, I build imagery by layering and reusing our existing AI-generated images alongside edits of those images. This collage approach ensures our final results are as far removed as possible from both the original training image data and initial AI renders, allowing us to develop unique ideas that align with our house style. Final images that appear on our platforms are also edited and customised to our specifications.
For our clients looking to plan ahead with AI imagery, I’d say identify the best tool that works for you, start honing your AI prompting skills and go on that journey to see if it benefits you. Looking ahead, I see WGSN AI imagery advancing the creative process from trend exploration to unique visual inspiration that can streamline client concepts and product development. With more efficient and sustainable idea-to-market cycles we aim to enable clients to blend ideas - from creating new catwalks looks to a new table lamp - to support innovative ideas in their own fields.
Change the way you work.
Imagine all the catwalk analysis you need - from live coverage to post-show analytics - in a single, dedicated place.
From analysing total looks and details from the runway to daily data updates on colour and category shifts, WGSN enables you to quickly identify emerging and evolving trends every season.
Discover the tools WGSN clients already have access to and how you can start translating catwalk inspiration and design direction into the right products.