I was curious as to exactly what settings get applied when "initialize" is run. Is it possible to createuse XMP files that I create myself in Lightroom, using images that already have adjustments?
In other words, is it completely required to initialize in LRT, or can I take a bunch of edited sequences in LR and just save them with the start and end key frames applied, and jump in at "reload" in LRT?
I know there are graduated and radial filters added, but these are often unnecessary; are there any other 'hidden' settings that happen with initialize?

Thanks in advance for any insight! I am just trying to better understand everything that is being done to the XMP data in LRT. I often find myself doing patchwork editing and want to know that I can come back to the middle of an incomplete workflow in the future. This could help save time for sure!

You can try it. With version 3.3 I made it a bit more tolerant, LRT will try to fix any "gaps" that Lightroom makes when writing XMP. Still I recommend the other way round: first initializing in LRT, then editing. But as I said: just try it.

BTW: You can save a snapshot of a whole sequence in LRT via the "Add Snapshot" function - this will "freeze" a current state of all metadata and let you restore later.