1.3.2

Bugfixes

• Remove torch hack and dependencies
• Run CI pipeline against macos-latest, ubuntu-latest, windows-latest and

1.3.1

Bugfixes

• Prevent creating of a face reference if --many-faces being used
• Fix invalid --frame-reference-number using the UI preview

1.3.0

Features

• Introduce --temp-frame-format to select between JPG and PNG
• Introduce --temp-frame-quality to reduce size of temporary frames
• Add support for video encoders h264_nvenc and hevc_nvenc
• Add --output prefix to --video-encoder and --video-quality
• Normalize quality arguments to a 0 to 100 range

Bugfixes

• Prevent processing if frames have not been extracted
• Enhance a larger box around the face to prevent unwanted borders

1.2.0

Features

• Lock faces to prevent bouncing via multiple faces
• Switch reference face via UP/DOWN arrows in the preview
• Switch frames via LEFT/RIGHT arrows in the preview
• Open the UI with source and target path by passing -s/--source and -t/--target
• Rename CLI argument --keep-audio to --skip-audio
• Speed up face enhancer by processing only the face box
• Enable MPS backend for face enhancer under MacOS
• Default for --max-memory has been removed
• Extract and process only the fps that are targeted
• Drag and drop source and target paths to the UI

Bugfixes

• Increase preview reload due to improved model instantiation
• Fix the crash when the video frames are lower than the number of threads
• Fix for the fps mismatch that caused fast/slow motion output

1.1.0

Features

• Introduce Queue for dynamic thread processing
• Add thread lock to face analyser
• Add memory usage to progress bar
• Show error for crashed frame processors

Bugfixes

• Fix memory leaks due temporary variables for the garbage collector
• Limit tensorflow usage to 1GB VRAM
• Remove choices from --frame-processor for third party support

1.0.0

Initial release!