Video Stabilizer

A video stabilizer using Three Step Search, Frame Position Smoothing and Cropping.

Features

• Three Step Search/Sparse Optical Flow: Efficiently searches for the best matching block in a reference frame to stabilize the motion. In addition to the Three Step Search, it's also possible to use Sparse Optical Flow for motion estimation.
• Frame Position Smoothing: Reduces edge effects on the motion signal through Gaussian filtering and static padding, ensuring smoother transitions at the beginning and end of the video.
• Cropping: Adjusts the frame size to ensure that the stabilized video does not have any blank borders due to the movement adjustment.

Installation

Clone the repository, make your config and config_log_uvicorn ini files into python-core, navigate to the root project folder, then run:

$ docker-compose up -d

This command will build the Docker containers and start the services in the background.

To shut it down, run:

$ docker-compose down

Usage

Once the containers are up and running, visit http://localhost:4200 in your web browser. Here, you can choose and customize the parameters to optimally stabilize your video.

Note: The performance of stabilization may vary depending on the video. Experiment with different parameters for optimal results.

About

This Video Stabilizer is designed to mitigate the shakiness and unintended movements in videos, making them smoother and more visually appealing. It utilizes either a Three Step Search algorithm or Sparse Optical Flow for motion estimation, Frame Position Smoothing to remove irregularities in motion, and Cropping to ensure the video frames are consistently sized. Additionally, static padding is employed to reduce edge effects on the motion signal, ensuring smoother transitions at the beginning and end of the video.