I stole and modified A113's recreation to do this since the 3ds spine shows the original vs me dicking around track. So I thank him making this example possible.
Doing what I did initially increased the G's at the bottom of the drop by a range of 0.2 to 0.3 (To around 4.5 on my computer) and then the y reduced by around 0.6 to 0.8 until the airtime started on the hill. My dicking around killed the G's a fair amount while still keeping them intact and maybe (seriously doubt it) they could reuse the same footers, and I still kept the "bump" at the top of the hill for airtime. They could raise it even further and ruin the airtime over the crest in the front to reduce G's even further.
Also regarding my G readings: My computer isn't the best, and I didn't adjust banking properly since this was very rough but conservatively I can say you can reduce the G's by 0.5 overall through the duration of the turn by doing what I did. Proper banking may bump the number down (highest reduction I got was 0.8, so that is why I'm conservatively going with 0.5 in reduction)
Image Insert: 318.46 KB