Wondering if it would be possible to use ffmpeg w/ v4l2 support (on say Ubuntu Linux) to create a virtual webcam device, that presents a transformed version of real webcam's output. The idea would be that the virtual webcam device could then be used by popular video conferencing applications like Teams, Zoom, Google Talk etc.
The nature of transformations I am looking at is cropping on few of the edges, applying overlay text, applying chroma-keying, background replacement etc. I am looking for a mechanism that has least system overhead as well, which is why considering ffmpeg.