Pompei

What Pompei is

Pompei is a free Python package for creating photo mosaics from movie frames. It extracts frames from a video, resizes them into thumbnails, and saves them with related information in a directory. For a mosaic, the package matches image regions to movie-frame thumbnails and assembles selected thumbnails into a PNG file. Options control mosaic width, how often a thumbnail can be reused, iteration count, and the time and spatial gaps between source frames. An experimental candidate-selection approach uses Scikit-learn to identify thumbnails that may suit a mosaic; MoviePy is installed as a dependency, while Scikit-learn is needed for that approach. The project documentation describes modular code and demonstrates customization by subclassing MovieFrames and overriding a method. The command-line API has not been implemented, so the package is used through Python rather than a finished command-line interface. Pompei runs on Linux and can be installed with pip or from source. It is open source under the MIT licence, and the project page invites commit proposals and requests for help on GitHub. PyPI lists version 0.1.04, released January 29, 2015.

Who it suits

Pompei suits Python users on Linux who want to create photo mosaics from movie frames and adjust thumbnail selection or spacing. Its customization approach may also suit developers comfortable subclassing its code.

Limits

The command-line API is not yet implemented. The experimental candidate-selection method requires Scikit-learn; MoviePy is installed as a dependency.

Getting started

1. Install the package with pip or from its source code.
2. Use Python to extract and store resized video-frame thumbnails.
3. Set mosaic width, thumbnail reuse, iteration count, and source-frame spacing.
4. Optionally use the experimental candidate-selection method, which requires Scikit-learn.
5. Customize behavior by subclassing MovieFrames and overriding a method.

Verdict

Pompei offers configurable movie-frame mosaics as a free, open-source Python package. Its lack of a command-line interface and the need to use Python are important considerations.