Look, as far as I remember, I used Finale quite a few years ago (was it Version 5 or maybe 3 ???) along with Smartscore Lite from Musitek.
I don't know if you are familiar with the product, but it allows you to scan a musical score (typically represented in a PDF file) and input the result into Finale. In practice, Smartscore is like an OCR reader, except that instead of characters, it recognizes notational objects (Notes, ornaments, clefs, etc...).
The goal I set for myself back then was to produce, in my spare time, modern digital scores that were not copyrighted but were "Open Source".
A project of this kind is MUTOPIA, which is now restarting after a year of inactivity and releases musical scores (also in digital form if you want to make your own editions, for example with Lylipond), generally under a Creative CommonsLicense.
Unfortunately, Mutopia's production pace is very low because photocopying is much faster than digitizing, as even a good analog-to-digital conversion program still requires a verification phase (and eventual correction) by a human, which is much slower. For this reason as well, IMSLP (the Petrucci Library) constantly increases its advantage over the previous project.
Returning to Finale, what I can tell you is that input and/or corrections were extremely fast once you learned the tricks of the trade and the shortcuts; the print quality and notation possibilities were equally remarkable.
A bit less successful was the suite of programs used for the other acquisition task I performed, consisting of MIDI acquisition aimed at score production.
Two difficulties: the first was that audio-to-MIDI converters, as is also the case today, are practically terrible as soon as you play something complex for them; the second was that even with an original MIDI file, as soon as it was authentically "expressive," the programs would crash.
This is because they lacked (as they still do today) criteria to transform musical event information into musical notation (Example: if three notes in the bass channel, separated by thirds, start in rapid succession within the span of an octave, represent the whole as a chord preceded by a notation sign; or: if for two measures the velocity increases progressively and for two it decreases progressively, add dynamic indications; etc. etc.). Keep in mind also that these criteria are not even absolute, but must be customized by whoever produces the score.
A good candidate, from this point of view, was Cakewalk, which had a programming language (the infamous CAL Cakewalk Algorithmic Language!) that, by allowing the manipulation of MIDI musical events, could recognize these things! It's a pity the language was not very powerful and did not allow for the generation of notational information, but only for modifying MIDI musical events!
I hope that, in the meantime, things have progressed, but I have the feeling that more investment has been made in production systems than in notational ones (business needs and publisher strategies agreeing!).
Bye and good luck with your work