OK, first lets keep shading and other aspects of depth illusion out of the discusion because while it does affect the perceived depth it does not relate to the science of the optics for creating a stereoscopic illusion. These other qualities of the image relate more to depth illusion on a 2D image and only enhance the depth of a stereo image.
The practice of sizing the optics for optimizing 3D stereo illusion is not subjective at all. It is a rather precise set of mathematical relationships that optimize the appearance for what you want to see in the stereo illusion. These have all been well defined in various literature and put in layman's terms by one Bernard Mendiburu. In his discussion of the physics he explains that the interocular distance of 65mm is not a limiting factor because the fact that we are not displaying the captured image in real size, but rather shrinking it to a display size. Therefore the use of greater than 65mm for our interaxial cameras is perfectly acceptable to minimize the flatness of distant subjects.
I mentioned the use of 18mm as the focal length of the camera lens but that is not representative of what the human eye is either. The focal length of the human eye has never been matched with lens optics. Peripheral vision that is near 180° while no horizontal line distortion and a front distance that mimics a 50° lens for front view distance has not yet been achieved. So, because of that no ultra wide angle lens has ever been produced. We can get a fisheye 6mm but then the image is farther way and severely curved.
So, in stereoscopic 3D the math dictates that the wider the the view angle the less the 3D effect as subject is pushed back. But keeping the same IA, the more we zoom in the better the 3D effect and closer to reality the image looks except that we lose scene width and object or subject depth in the distance. The latter may be recovered by using wider IA. The math says that the wider the field of view, the wider you need to have the IA to achieve the same depth of the distant object.
In the practical example of the water fall you zoom in and demonstrate the validity of the science because the water fall is now seen in front of the rock wall while in the wider shot the water fall is almost on the rock wall. What I am saying is by increasing the IA of the cameras well beyond the SBE you are limited to, you could achieve an appropriate spread of the water fall over the rock face and do that with a wider angle of view that would be closer to what you see with your eyes in the real world as you project it to your small screen at home.
In other words, you buy an SBE to improve the depth of your camera for distance but that is not the limit of the science. You can achieve more depth over greater distances with wider angle lenses by using greater spread than the SBE. You have nearly all the tools. You have GoPro's with very wide angle lenses. You have two of them. Now just mount them on a precise slide table to achieve the depth at distance.
My widest angle lens system for My twin system is 8.8mm on the DSLR's and I have a 1 Meter bench. The trick in using this maximum system is finding a location where I can use it and keep near objects from the scene as the near objects violate the stereo convergence. Places like the rim out on a point at the Grand Canyon would be one of the few locations this can be used. Next year I want to try to shoot NYC skyline in 3D with this setup. I have it now with a 18mm lens and 150mm by shooting from a high point off a cruise ship top deck in the center of NY Harbor. I have a feeling that the Yosemite El Capitan is too compromised with near objects to use the techniques I mentioned. This is one reason my bench uses the twin cameras at either end and the Z10K in the center for a tighter shot. My twin 3D system is not capable of zooming in sync.
PS- the water fall would also appear blurry due to the problem of the rapid motion not being genlocked, however, the mountain would be sharp as a tack. To fix this I would take your genlock cable for the GoPro's and extend it to 2 meters in length. Might buy an extra one as it would no longer fit in your housing. Just cut the genlock GoPro cable in two, match the wires up with some stock multi wire cable and splice it in. There are wiring diagrams on line I have seen for doing this, but you really don't need them.
One day we will get out there. I'm pretty much done with everything I wanted to do in Death Valley now. I may make it out there in April if I don't go on a cruise in May. We've also been looking at an Alaskan trip next summer too. All depends on the money. I like paying for the trips when I sign up so that is not an issue when it's time to go.
Ok enough for now. I need to get back to work on my video.