OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Image & Video AI

Ganimation

Ganimation is an open-source PyTorch research implementation for animating facial expressions from a single image. It conditions a generative adversarial network on facial Action Unit annotations, allowing users to control the magnitude of individual facial movements and combine them.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Ganimation — OpenViST project profile in Image & Video AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Ganimation when you need a research codebase for experimenting with facial-expression animation from a still image and can prepare the required image and Action Unit data.

Who it is for

  • Researchers studying facial-expression synthesis or GAN-based animation.
  • Developers exploring an Action Unit-conditioned image-to-animation pipeline.

Documented uses

  • Prepare image and Action Unit annotation data for the project’s training and test structure.
  • Train the documented model pipeline on a prepared dataset.
  • Test the implementation with an input image.
  • Explore how Action Unit activation strengths and combinations condition facial movements.
Before you choose

Requirements & limits

The project is a public GitHub repository identified as GPL-3.0 licensed.

  • The documented prerequisite is PyTorch 0.3.1, with Torch Vision and the project’s requirements.
  • The expected dataset includes image files, OpenFace-derived AU annotations, and separate training and test ID lists.
  • The repository documents scripts for training and testing but describes a code-based research workflow rather than a hosted interface.
  • The README requests citation when using the code or ideas from the paper.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare Action Unit annotations for a dataset

Create the AU annotation file used by GANimation’s dataset structure.

Before you start

  • A dataset directory with images and per-image Action Unit CSV files.
  • OpenFace for extracting Action Unit annotations.
  • PyTorch 0.3.1, Torch Vision and the dependencies listed by the project.

Steps

  1. Extract Action Units from each image with OpenFace.

  2. Save each image’s output as a CSV named to match the image.

  3. Run the command below to generate aus_openface.pkl.

    Documented command
    python data/prepare_au_annotations.py

  4. Include the imgs/ folder, generated annotation file, and train_ids.csv and test_ids.csv in the dataset directory.

Expected outcome

A dataset directory containing the image files, AU annotation dictionary and training/test image lists required by the project.

Guide limitations

  • The repository’s documented data structure requires both training and test ID files.
  • The guide covers data preparation; it does not verify model training or image output.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
albertpumarola/GANimation · Official homepage title · retained in its source language
How we review sourcesSuggest a correction