First thought based on comments...
Use differential backups every, say, 6 hours, to reduce the size/time of backup + FTP. Then reduce your full backup + FTP to weekends only. This avoids complexity of log shipping, simple to do, and only adds slight complexity to DR
I feel that differential backups are overlooked... I've suggested using them before:
Edit: after jcolebrand's comment I'll attempt to explain more
A differential backup only takes pages that have changed. Outside of any index maintenance (which can affect a lot of the database), only a few % of pages will change during a day. So a differential backup is a lot smaller than a full backup before any compression.
If you have a full backup, say weekly, you can then do daily differentials and ship them off site. A daily full backup with differentials will still require both files off site.
This should solve the problem of getting data from A to B, C and D quickly.
You probably need to restore both the full and latest differential to get the latest data but you can maybe work around this with NORECOVERY and a STANDBY file (I haven't tried it with a diff restore for years since I was last in a pure DBA job).
An added bonus is that diff backups are unrelated to ongoing log backups so you can separate any High Availability/DR requirement from the "get data to the code monkeys" requirement.
I see some issues if you have daily full backups by policy or audit, but the diff restore can be applied before any log restores to shorten recovery time. Unlike backups, diff and log restores do interact.
Hope I've covered most bases...
As you can see in this table, there is no conflict between Full backup and Log backup. They can occur at same time, and, when log backup completes, log will not be cleaned until full backup (or diff backup) completes.
Not all administrative tasks are allowed to run concurrently. In the table below, a black circle indicates two operations that cannot run in a database at the same time.
References:
Concurrent Administrative Operations
Best Answer
You are asking how Database Snapshots work - there is a good article on this on MS TechNet
The short answer: it is so fast because when you do a snapshot you do not backup or copy any data. All that is done - is an empty file created.