v1.0.1

Changed

• Face selection is now chunked, to improve performance when more than 1.000 images are processed.

Fixed

• Fixed a bug where face_selection_overrides could not be found in Windows because global variables do not work well with concurrency.
• Fixed exception due to incorrect usage of drawing function when multiple faces are detected in a single image.

v1.0.0

Initial release of Facemation.

Pre-processes images, normalizes images w.r.t. the face, adds captions, and runs FFmpeg on the output. Results are cached to speed up subsequent runs. Check the README for usage instructions.

Changes since v1.0.0-RC4

• Configuration validation gives clear errors when the configuration is invalid.
• Configuration for custom options with FFmpeg.
• Configurable path for FFmpeg.
• Numbers and strings can be used interchangeably in the configuration.
• Hugely enhanced README with clearer usage instructions and examples.
• Separated requirements.txt for different environments.
• Statically-linked libc for Linux, and statically-linked C++ DLLs for Windows.
• Missing third-party licenses have been added.

v1.0.0-RC4

Fix typing issue with PIL.

v1.0.0-RC3

• Remove OpenCV dependency.
• Use lower font thickness for captions.
• Add build scripts for easier building.
• Make README easier to understand.

v1.0.0-RC2

• A major overhaul of code structure to allow for easier extension in the future.
• Support for non-JPG images.
• Mostly-tested support for Windows.
• Significantly reduced binary size by using 5-point image model instead of 68-point model.

v1.0.0-RC1

First pre-release of Facemation, for testing the build process end-to-end and to identify remaining usability issues.