Can VMD Content Be Created Using AI?
While managing VMD in offline stores, we often encountered limitations in producing diverse content due to restricted budgets and time.
We thought, ‘If only we could bring the visuals in our minds to life, we could explore so many more themes! Could AI make that possible?’
That curiosity is what sparked this project—and here is the story that followed.
We thought, ‘If only we could bring the visuals in our minds to life, we could explore so many more themes! Could AI make that possible?’
That curiosity is what sparked this project—and here is the story that followed.
STEP 1. Theme Setting and Direction
Considering the characteristics of a retail store where customers over 60 make up the primary demographic, we determined that a storytelling-style VMD featuring characters of a similar age would foster empathy among visitors and support in-store counseling.
Rather than solely emphasizing anti-aging, we aimed to highlight ways to live more healthily and beautifully in everyday life—paired with self-manageable products.
As a result, the 2025 New Year season VMD theme was set as ‘Slow Aging.’
STEP 2. Content Creation with ChatGPT
Using GPT, we identified the preferences, characteristics, and expectations of customers who visit our store.
Based on these insights, we extracted prompts to generate POP copy that would appeal to the target audience through a slow aging daily routine, as well as prompts for creating diverse character visuals.
Although the information gathered through GPT still required further review, it was a valuable tool in drastically reducing the time and effort needed to search for vast amounts of data and distill key points.
▲ Step 1. Content Obtained Through ChatGPT Conversations
▲ Step 2. Refining the Content Extracted from GPT and Organizing it into VMD Copy
STEP 3. Creating Characters
During the project, we realized that empathy toward aging applies not only to customers over 60 but to people of all ages.
This led us to create a diverse range of characters representing various age groups, aiming to resonate with a broader audience and depict relatable daily life scenarios.
Using the slow-aging daily routines derived from GPT, we generated hundreds of images with Midjourney, focusing on realistic depictions of Korean individuals one might see in everyday life.
Initially, when we were still unfamiliar with AI tools, the visuals often featured awkward details like unnatural skin textures or hairstyles.
However, as we refined our understanding of prompt design and Midjourney’s capabilities, we were able to consistently produce characters that matched our vision.
▲ Early work: Character cuts with awkward expressions and hairstyles that were not usable
▲ Later work: After studying prompts and utilizing various features of Midjourney, the character cuts became more natural and realistic
Character cuts require significant time and cost—from model casting to photography and post-processing—making them difficult to use for VMD that is quickly displayed and taken down.
However, through this project, we were able to generate a large pool of models without the pressure of production costs and selectively choose only the cuts needed for VMD.
This was a highly meaningful experience.
STEP 4. Creating Images for the Store
While Midjourney enabled us to generate the desired character visuals, further adjustments were necessary for VMD application—such as enhancing resolution and refining the models’ styling details.
To utilize the generated model cuts in VMD assets like brochures, leaflets, and videos, we employed Photoshop AI to modify styling and backgrounds.
For the key visuals displayed on in-store columns, we used Magnific to upscale the resolution, allowing the images to be used as large-format visuals.
Finally, we turned the completed images into videos using Hailuo and paired them with GPT-generated copy to create a looping display within the pop-up space.
This approach allowed for a comprehensive and efficient pop-up space design using multiple AI tools.
▲ Midjourney → Photoshop AI: Changed the background of AI-extracted character images using Photoshop AI, received multiple suggestions, and selected the most natural-looking visual.
▲ Step 1. Initial image obtained from Midjourney
▲ Step 2. Used Photoshop AI to expand the background for a zoomed-out view, enhanced image resolution using Magnific, and color-corrected for store-wide visual tone.
▲ Step 1. Initial image obtained from Midjourney
▲ Step 2. Changed hair color, clothing, and background using Photoshop AI
▲ Step 3. Produced video content using Hailuo
Final
▲ Final video content for store display
▲ Final VMD content displayed in retail stores: brochures, videos, VMD copy, wide-color displays
Conclusion
This project, which began with some skepticism about whether AI could truly be used to create high-quality content,
turned out to be a journey of patience and persistence. While it initially seemed easy, achieving the exact visuals
imagined in our minds required significant time and trial. Still, by effectively leveraging AI’s ability to broaden the
design spectrum with just a few clicks, we found it to be a powerful tool for enhancing workflow efficiency.
turned out to be a journey of patience and persistence. While it initially seemed easy, achieving the exact visuals
imagined in our minds required significant time and trial. Still, by effectively leveraging AI’s ability to broaden the
design spectrum with just a few clicks, we found it to be a powerful tool for enhancing workflow efficiency.
To wrap up, we’d like to conclude this project with the video of the male model
that brought the biggest smiles to our team throughout the process.
that brought the biggest smiles to our team throughout the process.
- Amorepacific Creatives
- 기획 및 디자인
- 박문경, 임보라
