Everyone is fake - AI video test, no audio
Some more tests, a couple months back i was using SVD to try and get realistic webcam footage, the quality was low resolution and there were many artifacts, a few months later and the process has improved significantly, this uses Luma dream machine, there is a long way to go, especially with hands! but significant progress in realism has been achieved. Can't wait to start seeing multi-model models that include audio in the output generations.
