AI clothing generation tool

DressCode is a text-driven 3D clothing generation framework that aims to democratize design for novices, offering huge potential for fashion design, virtual try-on, and digital human creation. It first introduces SewingGPT, a GPT-based architecture that integrates cross-attention and text-conditional embeddings to generate sewing patterns with text guidance. It then customized a pre-trained Stable Diffusion to generate tile-based Physically Based Rendering (PBR) textures. By leveraging large language models, the framework generates CG-friendly garments through natural language interaction and also facilitates pattern completion and texture editing, streamlining the design process through user-friendly interactions.

ClotheDreamer is a 3D Gaussian-based text-guided clothing generation model that can generate high-fidelity, wearable 3D clothing assets from text descriptions. It uses a novel representation method, Disentangled Clothe Gaussian Splatting (DCGS), which allows clothing and human bodies to be optimized separately. This technology improves the quality of clothing and human body rendering through two-way Score Distillation Sampling (SDS) and supports custom clothing template input. ClotheDreamer's synthetic 3D garments can be easily applied to virtual try-ons and support physically accurate animations.