Something to consider is that we do provide a checkbox for switching automatic backups off for individual projects (in the Project ▸ Project Settings... window, under “Backups”). While there are a number of reasons you might use that, for this reason: large projects can be slow to back up and so sometimes it better to manage your own backups—typically with something that works automatically and in the background, such as nightly, so it doesn’t get in the way of working. Just don’t skip setting up that part of the equation, if you do choose to switch them off for this project (and really, you should have something like that anyway, at the least, on top of any Scrivener backups).
Also try turning off the zip compression option in the main Backup options tab. That adds a lot overhead. One downside is that your backup folder will be a lot bigger, but drive space is cheap—the bigger downside is that it is possible to accidentally live load a backup. Just be careful with them if you use that, and always create full copies in File Explorer, to another location, when looking through backup copies.
Also, the post you mention confuses me. For me, the zip backup takes much less time than the ‘normal’ (non-zip) backup (which is the one that’s taking a very long time).
So to clarify, there is some confusion there. Duplicating files across to a different folder is way faster than duplicating the files and then zip archiving them.
I have tried doing this and yes, it does indeed load quite a few files on startup. From the list it shows me, the vast majority are not files that I am currently using or displaying, is that normal?
I have noticed that some conditions can cause it to load too much on launch, but I’ve never managed to pin down what causes that. When doing very focused tests I always get what should be the correct result: only loading what is in an editor somewhere. For example if I have “Text A” in the main editor, and the inspector open to the Bookmarks tab, which is showing “Text B” in the lower half of the inspector, then on reload I see the RTF files for both of these files being loaded. Makes sense—it needs to show them. If I add “Text C” to the Bookmark list and leave it unselected, I still get the same result.
That’s just one example, there are so many things that might cause excessive and unwarranted loading though, that testing everything and all of the various combinations in between can be pretty time-consuming. If you spot anything in your project window layout that might be a clue, let me know.
Oh, I was referring to the screenshots as a function, the one that tracks the different versions of the documents.
Ah, Snapshots you mean. Those shouldn’t be of impact. Even if you happen to leave one selected in the inspector when you close the project, it doesn’t remember that on reload and won’t even load it. That would be just one single snapshot though.
Excessive snapshots will slow down backups though, as they are all individual files. Considering taking a full named backup (File ▸ Back Up ▸ Back Up To...) that indicates it has all your historic snapshots, and then go into Documents ▸ Snapshots ▸ Snapshots Manager and run a date search. For example, <1y is everything older than a year. Then you can select them all and click the - button at the bottom to bulk delete them. This is why you want to take a full backup that is set aside from the normal backup rotation, first. 
, although the backup still takes a while.
