I think your second option is better. Carry the baby around and just mask you out. Just make sure the camera doesn't move and your hand doesn't occlude the baby at all. Luckily, they're usually a little chubby, so that should help. Alternatively, you can make sure the baby's wardrobe helps to hide the holding hand.
If you lock down the camera and shoot a clean plate then it should be easy to Roto you out of the frame.
If you want to use a handheld camera then I would use a green screen. Expect to take a lot more time making a handheld shot look right.
Just for clarification:
- I place the camera on the tripod and shoot some video of the environment
- I step into the shot with the baby and try to minimize how much of me is in the shot
- I roto the part of me that is in the shot
- With both clips, the part of me that was removed will be replaced by the background from the other shot
1. Do these steps sound correct?
2. After rotoscoping me out of the image, what exactly am I doing with the masked out portion, that part I am not clear on.
If you're masked out, just plop the shot with nobody in it underneath.
It doesn't matter how much of you is in the shot. What matters most is that you don't occlude the baby.
Basically, you're cutting the baby out and placing it in the scene without you. We're just trying to help you have less rotoscoping/cloning work.