Easy way -- shoot in 4k, use Track Motion and keyframes in a 1080p project for your punch zooms.

Really? They actually do this by moving a camera or with 2 different cameras? This is so common it seems crazy you'd have to do that. Is there a reason you wouldn't do it just with Event Pan/Crop and zoom after the fact? Also, if doing it that way, how do you keep the audio narration sounding consistent? It would seem that if switching the narration from one shoot to another would cause inconsistency.

In the past you wouldn't zoom in with software because the resolution would not hold up. As Musicvid say, 4k would allow better quality at zooming in a 1080p project. But I guarantee you they shot this one with two camera angles. Very common, not hard to do. Keep narration sounding consistent by using a real microphone either on a boom or lavalier. Not an on camera mic.

Easy way -- shoot in 4k, use Track Motion and keyframes in a 1080p project for your punch zooms.

Thanks. What if I only have a Canon Vixia and can't shoot in 4K? Can I still do this using Track Motion and keyframes? Also, I don't suppose there is a tutorial on this somewhere you might be aware of?

If you use track motion, you will see a degradation in quality.

In the past you wouldn't zoom in with software because the resolution would not hold up. As Musicvid say, 4k would allow better quality at zooming in a 1080p project. But I guarantee you they shot this one with two camera angles. Very common, not hard to do. Keep narration sounding consistent by using a real microphone either on a boom or lavalier. Not an on camera mic.

Wow. Much more work than I thought this was goin to be. Thanks for the clarification. And I am using a separate mic recording to my computer. So I guess as long as I'm standing in the same place for both shoots, it should be fine. Thanks again!

If you use track motion, you will see a degradation in quality.

No, not in a 1080 output. Its more-or less SOP.

Youtubers are generally not using multiple cameras or takes.

Track motion does not work in original resolution.(or at least it didn't used to) You would need to use Pan/Crop. and I was referring to using it with the Vixia camera which is probably 1920 x 1080.
(musicvid changed his post after I made this comment, so my response doesn't coincide with his previous comment)

Yes, Track Motion processes based on the project size, Pan/Crop processes on base of the source media size.

Yes, my cold made me do it. Note to self: stay off the streets today.

(Overpost was unintentional).

The only way you'd want to zoom in on 1080p video is if your final render resolution was below that, like 720 or even 480p. However, your video will look blurry because even our phones have 1080p screens these days, and you probably won't get enough extra pixels by only dropping to 720p, which gives you roughly 50% of your pixels you can zoom in on without losing resolution.

You can use sharpening on zoomed in shots to give the illusion of the image not being soft but sharpening people's faces makes them look unhealthy and unattractive

Thanks everyone. I did a few experiments today and I think the degradation for zooming in is not that bad for what I'm doing.