I think capturing their personalities was the easy part. While trying to make the maps and figure out the basic setup, I listened to like 30 hours of footage from the original streams, so their speech patterns were pretty easy to recall at that point.