FashionRNA: Interactive Reimagination of Fashion Heritage

Haneol Kim, JONGHWAN BAE, Seonwoo Shin, Sanghun Park

Keywords: Fashion Generation, Extended Reality, Multimodal Interaction, PBR Material Generation, Digital Heritage

TL;DR: FashionRNA is an AI platform that uses voice and visual inputs to reinterpret historical garments, generating 3D models with PBR materials for dynamic visualization.

Abstract: In this paper, we propose a novel framework that integrates physical fashion heritage with interactive digital creation, facilitating the reinterpretation of historical garments via an intuitive multimodal interface. This interface allows users to manipulate digital representations of historical artifacts, specifying stylistic modifications through a combination of visual and vocal inputs. These requests are processed by a computational pipeline comprising state-of-the-art artificial intelligence (AI) models. The pipeline's workflow is as follows: it first retrieves a target style based on user input, then generates a corresponding 3D garment model, and concurrently synthesizes Physically Based Rendering (PBR) materials derived from the source pattern. The resulting 3D asset is optimized for real-time rendering, enabling immediate visualization. The results demonstrate that this approach generates high-fidelity and stylistically coherent outcomes, effectively transforming static cultural artifacts into dynamic, interactive digital assets.

Submission Number: 9