Home
Home
In the original campaign, Vedett allowed users not only to upload a photo but also to add a title to their label: users could add a word after 'extra' in Vedett’s house style. Vedett’s website is full of the funniest results from this action:
This combination of word and image inspired us: could we leverage both elements as input for an AI model, rendering the label even more personalized?
In the AI-enhanced model, the user experience was essentially unchanged: users upload a photo and select a word to appear next to 'extra' as the title. However, instead of just slapping on a basic black-and-white filter on the image, we sprinkled in some AI magic to craft the perfect label.
When a user uploads a photo, a GPT model kicks into gear, describing the individual and their recognizable features—hairstyle, accessories, attire. Subsequently, DALL-E generates an image of the individual, using the GPT model's description and the chosen 'Extra' word as input. For instance, you can see in the example that the AI whips up an image of a seasoned DIYer, based solely on the acronym ‘DIY’.
In the final stage, a face swap is performed, merging the uploaded photo with the generated image. The results illustrate how the Vedett labels have become even more personalized.
Vedett might not have gone through with this campaign in the end — we built this demo mainly out of our own interest. However, the model's fantastic results could offer a great chance to boost interactivity on their website, enhance customer engagement, and increase brand awareness.
Fueled by the creativity and passion of our technologist Stijn, this project shows just how far technology can go when paired with our bold ideas. It proves that its possibilities are only as limited as our imagination, spanning practically every sector of daily life.
At nocomputer, we love digging into all these possibilities. Stay tuned for more projects like this!