Menu
Bringing a custom product to life for fashion brands and custom clothing companies are very expensive and time-consuming.
At a custom bridal gown company, illustrators were shown a Pinterest board of dresses and asked to design and draw a custom dress based on the commonalities between dozens of images. Often times, the bride would want to choose between many different designs, which results in increased costs and longer timelines. The business wanted to reduce the time and cost spent creating the ideal gown for each client.
Upon reviewing the data and their existing software that enabled brides to create the Pinterest board, we consulted on prototyping two types of machine learning models: a Neural Style Transfer model and a GANS model. The Neural style transfer allowed the bride to choose a shaped dress such as Mermaid style, and then it generated the contents of this shape based off of the commonalities in the bride’s selected images. The GANS model processed the bride’s selected images and generated a new dress based off of these images. We then refined the solutions based on the accuracy and integration potential within the company’s existing processes.