AI for Fashion Design: Writing Effective Prompts with Midjourney

#Ai #Tools #Midjourney #Fashion

Hello Fashion Designers,

In this post for beginners working with the Midjourney tool and generative AI for fashion design, we share the foundational principles of writing text prompts and provide examples of simulations. Let’s take a look at this very short simple example:

"Prompt: /imagine black swan and mannequin in red drapes MoMa stage, eloquence --v 6.1."

Let's break down why it works so well. When requesting an image from any generative AI, it's crucial to understand how the text is interpreted — the shorter the text, the more creative freedom you allow for the AI. A more detailed prompt, longer than 25 words, sets restictions by limiting the number of images from the AI's library. Let's not forget the fundamental purpose of these tools; they are neither to execute an illustration like Photoshop or a pencil does, nor to replicate an iconic look from a vintage fashion magazine.

Generative AI crafts the image pixel by pixel, mathematically, while visualizing ideas directly from your imagination.

When designing fashion with AI, start by thinking about shape, color, and texture, the same way you would drape on a mannequin. Describe only the key elements and crucial details; at the end, set the ambiance, aesthetic, and camera settings if you are familiar with photography.

Consider how the listed elements will combine their visual qualities and which concepts you would like to assign more weight. In the provided example, most elements lean towards dramatism — black vs. red, drapes and stage are linked to the opera, hence the red curtain and heavy textures. The mannequin and swan represent elegance and sleek silhouettes, textures, and overall presence. The swan being black signifies rarity. The setting chosen is MoMa, allowing us to tap into artistic expression without providing an explicit list of references. Finally, the word "eloquence" is a nod to classici but outdated literary language, which, while bringing no particular shapes or colors, ties up the entire prompt like glue, allowing AI to select the missing image cues that would bond the outcomes harmoniously.

Our prompt is concise for a reason; we allow high creativity from the tool while providing the key elements. This type of language, structured on the visual weight of elements, is a way to instruct the machine:

"Please draft a dramatic scene that features a performance from an unknown opera, include elegant elements, particularly the swan, present the glossy mannequin figure in a garment of your choice inspired by its environment, adjust the light setting and the volume of the space according to an overall fantasy, draft the shapes and textures inspired by modern art exhibitions, tie it all up harmoniously and fill in the blanks relying on your intuition and dataset."

When testing the results, usually 30 seconds and 4 options are provided. Adjust the prompt accordingly, introducing new words one by one while assessing how the gravity of the text is changing the evolution of your designs.

Do not rely on the magic of sample prompts and examples available online. Instead, try to mindfully understand the fundamental principles of how to convert your own language into a successful formula for a machine. Many novel apps exist for image generation, tailored to diverse needs, usually for social media content or graphic design. However, there is not yet a tool dedicated solely to fashion design with AI. These tools are based on universal neural concepts and built on human brain examples. Start simply by engaging with existing ones that are already user-friendly, easy, and swift in use.

And finally remember, this is not an editorial photoshoot, no need to torture the tools that serves different purpose than Canon camere, develop your design concepts the way you work on the mannequin. Idea - Iterations - Refinement.

Don't hesitate to ask for advice or clarifications in DMs. A good start and understanding the fundamentals of Generative AI and neural graphic engines will allow you to filter tools and functions, to learn efficiently, and grow quickly.

Speak soon,

Sergiy

Atelier Karusel? Paris

Graduate of Central Saint Martins; Formerly at Louis Vuitton

#Fashion #Design #GenerativeAI #Innovation