Hi, I started this project with the idea to clone d-id and create an alternative. As you can see, the video generated is not as good as d-id, and currently looking for feedback and suggestions on what technology I can use to have better video output (natural head movement, etc.) or on what direction you think I should take this. I wrote a more detailed writeup of what I learned in this blogpost https://nonstop-lung-14d.notion.site/Launched-an-AI-Video-Av...