DressCode is a text-driven 3D garment generation framework that enables users to create high-quality garments through natural language interaction. The framework includes SewingGPT, a GPT-based model that generates sewing patterns based on text prompts, and a pre-trained Stable Diffusion model that generates PBR textures. SewingGPT converts sewing pattern parameters into a sequence of quantized tokens and uses a decoder-only Transformer with text-conditioned embeddings to generate token sequences. The generated sequences are then de-quantized to reconstruct the sewing patterns. The framework also allows for pattern completion and texture editing, streamlining the design process through user-friendly interaction. The generated garments can be seamlessly integrated with CG pipelines, supporting post-editing and animation while ensuring high-quality rendering. The framework has been evaluated against other state-of-the-art methods, demonstrating superior quality and alignment with input prompts. User studies further validate the high-quality rendering results, highlighting its practical utility and potential in production settings. The project page is available at https://IHe-KaiL.github.io/DressCode/. The framework's contributions include proposing a first text-driven garment generation pipeline with high-quality sewing patterns and physically-based textures, introducing a novel generative paradigm for sewing patterns as a sequence of tokens, and tailoring a diffusion model for vivid texture generation of garments from text prompts. The framework enables users to generate customized garments through natural language interaction, facilitating garment generation, completion, and editing. The framework has been tested on various datasets and has shown promising results in generating high-quality garments with diverse styles and textures. The framework's ability to generate garments with complex stitching relationships and multi-layered garments has been demonstrated, although there are limitations in generating garments outside the domain of the training dataset. The framework's potential applications include virtual try-on, fashion design, and digital human creation. The framework's effectiveness has been validated through experiments and user studies, demonstrating its superiority in producing CG-friendly garments that align closely with input prompts. The framework's contributions to the field of 3D garment generation are significant, offering a new approach to garment design that is accessible and interactive.DressCode is a text-driven 3D garment generation framework that enables users to create high-quality garments through natural language interaction. The framework includes SewingGPT, a GPT-based model that generates sewing patterns based on text prompts, and a pre-trained Stable Diffusion model that generates PBR textures. SewingGPT converts sewing pattern parameters into a sequence of quantized tokens and uses a decoder-only Transformer with text-conditioned embeddings to generate token sequences. The generated sequences are then de-quantized to reconstruct the sewing patterns. The framework also allows for pattern completion and texture editing, streamlining the design process through user-friendly interaction. The generated garments can be seamlessly integrated with CG pipelines, supporting post-editing and animation while ensuring high-quality rendering. The framework has been evaluated against other state-of-the-art methods, demonstrating superior quality and alignment with input prompts. User studies further validate the high-quality rendering results, highlighting its practical utility and potential in production settings. The project page is available at https://IHe-KaiL.github.io/DressCode/. The framework's contributions include proposing a first text-driven garment generation pipeline with high-quality sewing patterns and physically-based textures, introducing a novel generative paradigm for sewing patterns as a sequence of tokens, and tailoring a diffusion model for vivid texture generation of garments from text prompts. The framework enables users to generate customized garments through natural language interaction, facilitating garment generation, completion, and editing. The framework has been tested on various datasets and has shown promising results in generating high-quality garments with diverse styles and textures. The framework's ability to generate garments with complex stitching relationships and multi-layered garments has been demonstrated, although there are limitations in generating garments outside the domain of the training dataset. The framework's potential applications include virtual try-on, fashion design, and digital human creation. The framework's effectiveness has been validated through experiments and user studies, demonstrating its superiority in producing CG-friendly garments that align closely with input prompts. The framework's contributions to the field of 3D garment generation are significant, offering a new approach to garment design that is accessible and interactive.