I don’t care how the background is generated, if the making algorithm can’t understand which part is my head, which is the headphones and which is the headrest on my chair. Unless I use green screen, the auto detect for the background is too inaccurate do use any background filters.
How about AI-ing that instead?
I think about this all the time for autocorrect. The world’s most advanced AIs, built on text prediction models. My phone can’t tell the difference between bit and but in a sentence. Or ducking.
I’ve used Google and Zoom at work at different t companies. It’s surprising how Google lags behind Zoom in features. Google is worse at identifying background as background. And Zoom has had video backgrounds for a while now.
deleted by creator