# RameenAbdal/StyleFlow

StyleFlow: Attribute-conditioned Exploration of StyleGAN-generated Images using  Conditional Continuous Normalizing Flows (ACM TOG 2021)

Repository: https://github.com/RameenAbdal/StyleFlow
Canonical: https://ross.abutalabs.com/products/styleflow
Homepage: http://rameenabdal.github.io/StyleFlow
Language: Python
License Family: other
Topics: stylegan, sequential-edits, quality, faces
Last push: 2023-03-24T22:20:54+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2290, "days_push": 1258, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2442, forks 339 (observed 2026-08-28T04:06:52.440824+00:00)

## What it is
StyleFlow is a research implementation of attribute-conditioned exploration of StyleGAN-generated images using conditional continuous normalizing flows (ACM TOG 2021). It supports attribute-conditioned sampling and fine-grained, disentangled attribute-controlled editing of faces and cars in the StyleGAN latent space.

## Use cases
- edit attributes of a StyleGAN-generated face like pose, expression, age, or gender
- perform sequential disentangled edits on GAN latent codes without unwanted attribute changes
- sample new images conditioned on target attributes in the StyleGAN latent space
- project real photographs into the StyleGAN latent space and edit them
- reproduce results from the ACM TOG 2021 StyleFlow paper
- explore the car latent space of StyleGAN with attribute controls

## When to choose
- you need disentangled, attribute-controlled editing of StyleGAN latent codes
- you want a reference implementation of conditional continuous normalizing flows in GAN latent spaces
- you are doing research on GAN latent space exploration or controllable image synthesis

## When to avoid
- you need a production-ready or actively maintained tool with a license
- you want to edit images without a StyleGAN2 setup including custom TensorFlow ops and CUDA
- you need support for modern GPU stacks, since the code targets PyTorch 1.1 and TensorFlow 1.15
- you want general-purpose image editing outside GAN latent spaces

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, data-science
- domain: machine-learning, computer-vision, image-processing, artificial-intelligence
- platform: python
- tags: stylegan, normalizing-flows, gan-latent-space, face-editing, research-code, pytorch, tensorflow, disentangled-edits, attribute-conditioned-sampling, linux, gpu

## Member repositories
- RameenAbdal/StyleFlow (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:52.440824+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:30:37.097718+00:00, confidence not recorded.
  - readme: https://github.com/RameenAbdal/StyleFlow (fetched 2026-08-28T04:06:52.440824+00:00, sha e24bf7367442)
  - homepage: http://rameenabdal.github.io/StyleFlow (fetched 2026-08-29T10:12:07.165028+00:00, sha c1d4655c55d4)
- Data as of 2026-08-30T08:39:29.467469+00:00.