Backlog

Features

• Introduce face animation for static images

• Add bounding box and face landmark smoothing

• Generated steps for a job by detecting scenes

• Feature manipulation using interfacegan or styleganx

• Replace wave2lip with a high quality lip-syncer model

• Add FaRL and BiSeNet face parser models

• Introduce remote webcam via real-time streaming protocol

• Expose benchmark and webcam settings to the CLI

• Detect webcam devices via ffmpeg to make them selectable

• Save and load face embedding collection files

• Audio processor to mix the volume of the audio sources

• Support for DeepFaceLab models

• Detect the optimal configuration using the benchmark suite

• Replace gender_age with the high accuracy mivolo model

• Support for 180 and 360 degree videos

• Multi inference pool to support multiple devices

• Add GIF support

Bugs

• Fix retinaface not being able to detect cropped face