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

First Order Motion Model for Image Animation

First Order Motion Model for Image Animation is an MIT-licensed research model and source-code implementation that transfers motion from a driving video to an object in a source image, producing an animated sequence. Its repository includes a Python demo, dataset configurations, pretrained checkpoints, and code for training and video-reconstruction evaluation.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

First Order Motion Model for Image Animation — 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 the demo when you have a suitable source image, driving video, task-specific configuration and checkpoint, and want to generate an animation locally. Use the training or evaluation code when working with a dataset and model checkpoints.

Who it is for

  • Researchers studying image animation and motion transfer.
  • Developers building or adapting image-animation workflows.
  • Users who can prepare cropped image and video inputs and run Python code.

Documented uses

  • Animate an object in a still source image using motion from a driving video.
  • Train a model on a specific dataset and save checkpoints.
  • Generate video reconstructions or animate test-set examples for evaluation.
Before you choose

Requirements & limits

The implementation is publicly available on GitHub under an MIT license. The repository links to a Colab notebook and a Hugging Face Space as additional demo entry points.

  • Both source image and driving video should be cropped before use.
  • Configurations and separately trained networks are associated with different tasks; choose the matching configuration and checkpoint.
  • The documented local setup uses Python 3 and the dependencies in requirements.txt.
  • The Linux Docker option specifies Docker 19.03 or later and nvidia-docker.
  • The repository does not document a general-purpose API.

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

Run the image-animation demo

Use a pretrained checkpoint, source image and driving video with the repository's documented demo command.

Before you start

  • Python 3 and dependencies installed with pip install -r requirements.txt.
  • A dataset YAML configuration, pretrained checkpoint, source image and driving video.
  • Cropped source image and driving video.

Steps

  1. Install the repository dependencies with the command below.

    Documented command
    pip install -r requirements.txt

  2. Obtain a pretrained checkpoint and choose a configuration file for the dataset.

  3. Run the command below.

    Documented command
    python demo.py --config config/dataset_name.yaml --driving_video path/to/driving --source_image path/to/source --checkpoint path/to/checkpoint --relative --adapt_scale

  4. Retrieve the generated result.mp4.

Expected outcome

The demo writes an animated result to `result.mp4`, transferring motion from the driving video to the object in the source image.

Guide limitations

  • The README requires cropped inputs but does not specify a universal cropping method.
  • The project describes separate networks for different tasks; configuration and checkpoint selection depend on the task.
  • The README's crop-suggestion script requires the face-alignment library.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
AliaksandrSiarohin/first-order-model · Official homepage title · retained in its source language
How we review sourcesSuggest a correction