For a full backup we make a 100% copy of our data (all data) to an alternate location.
For an incremental backup we create little increments of changes. An example would be that we take a full backup on Sunday. Monday’s incremental backup will be the changes since Sundays full. Tuesday’s incremental backup is the changes since Mondays incremental. Wednesday’s incremental backup will be the changes since Tuesdays incremental. So we are building an incremental dependency. When we do a restore, we would have to restore the latest full and then all subsequent incrementals in sequence. So if, for instance, we lost Thursdays incremental backup, we couldn’t restore past Wednesday. So any changes on Thursday and Friday would be lost.
If we take a full backup on Sunday, Monday’s differential is the changes since Sunday. Tuesday’s differential is the changes since Sunday. Wednesday’s differential is the changes since Sunday. So the differential backups get larger. We don’t have a dependency so when we do a restore, we now restore the latest full and the latest differential. So if we lost Wednesday’s differential we could still restore Thursdays because we don’t need Monday, Tuesday, Wednesday (we have recaptured all of that data).