nvlabs/stylegan

★ 14,402PythonAudience · researcherComplexity · 4/5LicenseSetup · hard

Mindmap

mindmap
  root((StyleGAN))
    What it does
      Generates fake images
      Faces cars bedrooms cats
      Style-based control
    How it works
      Style transfer approach
      Separate coarse fine control
      Pre-trained models included
    Tech Stack
      Python
      TensorFlow 1.10
      NVIDIA GPU required
    License
      CC BY-NC only
      Non-commercial use
      NVIDIA attribution required

mindmap root((StyleGAN)) What it does Generates fake images Faces cars bedrooms cats Style-based control How it works Style transfer approach Separate coarse fine control Pre-trained models included Tech Stack Python TensorFlow 1.10 NVIDIA GPU required License CC BY-NC only Non-commercial use NVIDIA attribution required

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Generate photorealistic synthetic faces for research datasets or creative projects without using real people's likenesses.

USE CASE 2

Explore independent control over coarse features like face shape and fine details like hair strands in AI-generated images.

USE CASE 3

Use pre-trained models to generate bedroom, car, or cat images immediately without training a model from scratch.

Tech stack

PythonTensorFlow

Getting it running

Difficulty · hard Time to first run · 1day+

Requires TensorFlow 1.10 (an older version) and an NVIDIA GPU with CUDA, not compatible with modern TensorFlow without significant modification.

Free for non-commercial use with attribution, commercial use requires a separate license from NVIDIA.

In plain English

StyleGAN is the official code release from NVIDIA for a research paper published in 2018 that introduced a new way to train an AI to generate realistic images. The paper is titled "A Style-Based Generator Architecture for Generative Adversarial Networks." The teaser image in the README shows human faces that do not belong to real people. They were entirely synthesized by the model. The core idea behind StyleGAN is a different approach to how the image generator is structured. Traditional image generators start from a single random input and transform it step by step into an image. StyleGAN instead draws inspiration from "style transfer," a technique where the visual style of one image is applied to another. This architecture allows separate control over different aspects of the generated image: high-level traits like the overall shape and identity of a face are handled separately from fine details like freckles or individual hair strands. The result is that you can adjust these aspects at different scales without them interfering with each other. The repository provides pre-trained model files for generating faces (trained on two large face datasets), bedroom interiors, cars, and cats. You can load these pre-trained models and immediately generate new images without doing your own training. The code is written in Python and uses TensorFlow 1.10, which is an older version of the TensorFlow machine learning library. The material is released under a Creative Commons BY-NC license, meaning it is free for non-commercial use with proper attribution, but commercial use requires a separate license from NVIDIA. The paper was published on arXiv and an accompanying video demonstrating the results is also available. A newer version, StyleGAN2, has since been released and the README points to that repository for anyone wanting the more up-to-date implementation.

Copy-paste prompts

Prompt 1

I have the StyleGAN pre-trained face model. How do I generate a batch of 10 random face images and save them to disk using the provided Python scripts?

Prompt 2

Using StyleGAN's style-mixing feature, how do I combine the coarse structure of one generated face with the fine details of another?

Prompt 3

How do I load the StyleGAN bedroom pre-trained model and generate a grid of sample images to see what the model has learned?

Prompt 4

What does the style-based architecture mean in practice, how does StyleGAN give separate control over large-scale and small-scale image features?

Open on GitHub → Explain another repo

← nvlabs on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.