If you're using 3D anyway, make use of the depth. Not in terms of movement, but in terms of scener. If you're going to place the camera at a non ortographic angle, display shapes in the caves. If you're going to use the perspective of the camera and not render it as an ortographic one,use multiple image planes and add splits of the cave going into the depth. Unreachable, but logically apparant.
Also, you keep showing new stuff, and haven't gone back to adjust a single thing. If this is a showcase, make it good. If it isn't, show that you listen to critique. I know I sound blunt there

no offense intended but it's my most direct way of putting it, and I do mean it that you should take more time to stand still at what you are doing before making more progress. The more you do, the more you redo. That's a rule of thumb in any project

- My main critique for now: integrate your elements of choice logically into the design.