I'll try to avoid details, as much of that will be self-evident once you start using the program. I suggest you start with the beta (currently version 1.0.2.3), as it's stable and includes some useful new features.
You probably recall me stating at Interact that importing from PVD to MC is fully automatic. I'm referring to the process PvdImport provides, once properly configured. It monitors the PVD database directly, and automatically updates any records that have changed. The process of collecting data within PVD is better described as semi-automatic. That's not a bad thing. You're probably considering this because you've learned from painful experience the J River approach of trying to be fully automatic just gives you incomplete results for an arbitrary set of data that wasn't what you wanted in the first place. PVD gives you control over what data to collect and where to get it from.
The file scanning process—that matches files to records and creates a new record if one doesn't exist—uses a set of regex (processed in order, first match wins) to do the matching and extracting of data from pathnames (to create new records). If your file naming practice is 100% consistent, then the scanner—if properly configured—will be 100% successful. For practical purposes, expect it to be 95% successful, with the rest easily fixed manually. The exceptions will be mostly unavoidable issues with unusual or ambiguous video titles. Those will be minimized if both an accurate [Title] and [Year] are provided (in the pathname).
The scanner can be configured to run specified meta data import routines on the files it finds, but you still have to approve the processing of the batch. Generally, I prefer to data importing as a separate step I can oversee. Movies and episodes (at least) require different routines. I'd rather select the items I want updated, then run the appropriate updates. Those can be run in batches. For new movies, for example, I run one batch that combines data from four different sources.
So if your expectation is for something you can configure and forget, PVD is probably not for you. It's for those who care about the process of collecting data about movies (as any 'collector' would), and therefore want as much control as possible over it. If that is important to you, the result is a good one. Both MC and PVD will add files to their database as they 'appear' in scanned folders. In PVD, you run the import applicable import routines and make whatever other adjustments you want to make (e.g., adding a bit of data manually, finding a better poster, etc.). All this data then appears automatically in MC. The best approach is to do all the data maintenance in PVD, and then just enjoy it in MC.
