Voice To Virtual Try On
FIA’s Voice to Virtual Try-On turns speech into 3D garments for virtual try-on via Lens Studio.
Developed by FIA’s AI Intern Aslı Ilhan, this Lens Studio pipeline turns voice prompts into multi-view garment images, detailed meshes, and textured, try-on-ready digital clothing. By connecting speech recognition, image generation, and NVIDIA-powered reconstruction, it preserves the full intent of a designer’s imagination as it moves from words to wearable form.
From spoken vision to real-time virtual try-on workflows that simulate a designer’s creative flow.
Using Snapchat’s speech recognition and a custom Snap Mask prototype, the system converts verbal prompts directly into backend generation requests that produce multi-view garment images in real time. The speech-to-garment workflow demonstrates how designers can describe their vision aloud and immediately see it progress from voice, to code, to cloth, amplifying creative expression rather than replacing it.
PartPacker and NVIDIA: procedural, editable 3D garments optimised for cloth simulation and virtual try-on.
PartPacker, powered by NVIDIA AI, transforms 2D garment images into segmented, part-based 3D meshes that are fully editable, procedural, and optimised for cloth simulation. Each garment is broken into meaningful pieces to retain shape, allow fine-tuned texture work, and deliver wearable, try-on-ready clothes. PartPacker also generated the synthetic dataset used across the pipeline, making it essential to the project’s high-fidelity digital fashion workflow.
Explore new possibilities
Get in Touch