Skip to content

DressCode

Generate 3D garments from text descriptions

Visit official site

What DressCode is

DressCode is described as a text-driven 3D garment generation framework that first introduces SewingGPT, a GPT-based architecture using cross-attention and text condition embeddings to guide sewing pattern generation. The framework also customises a pre-trained Stable Diffusion to produce tile-based PBR textures.

What you can do with it

  • Generate garment patterns and textures through natural language interaction
  • Complete patterns and edit textures through a user-friendly interface
  • Produce CG-friendly garments with a large language model
  • Obtain high-quality renderings aligned with the input prompt

Application scenarios

  • Designers generate garment sketches and textures quickly
  • Fashion brands provide virtual try-on services
  • Digital content creators design clothing for virtual characters

Getting started

  • Clone or download the DressCode project to a local environment
  • Set environment variables and update local path configuration from the docs
  • Install required dependencies such as Stable Diffusion 2-1
  • Train the SewingGPT model with the provided scripts and dataset

Pros & cons

✓ What we like

  • Pattern completion and texture editing are supported
  • Output supports later simulation and animation in 3D garment software

! What to watch out for

  • Setup requires training the SewingGPT model with provided scripts
  • PBR texture generation depends on a pre-trained Stable Diffusion

FAQ

What does SewingGPT do?

It is a GPT-based architecture that uses cross-attention and text condition embeddings to guide sewing pattern generation.

What do I need to install?

The steps name dependencies such as Stable Diffusion 2-1.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review