I'm still not 100% sure whether this is right for me, but this VR teleport movement model works quite well.
I've being playing around with more directional voxel style lighting. There is a lamp in the house and the light is shining out the window.
I've being working on being able to build an actual house. The hard bit here is the roof. This is because the hexagons are not regular symmetric. The solution was to use the FBX sdk to generate the 7 different variants I need (including a pentagon version). This of course means I have a high initial cost to develop the code, but if I need to change something, it is quite quick. The code reads in mesh components from a fbx file and assembles them creating a ue4 ready fbx file that I can directly import.
In my quest to improve VR performance, I've taken a leaf out of the Hierarchical Instance Static Mesh hand book and put in code not to render tiles that you can't see. They still exist because they need to for collision purposes, but I don't to get the GPU to render them. I initially tried placing non visible tiles for half the sphere, then I went for more. I ended up getting that mine craft look when things magically appear.
Just watched the video on "the rendering below" and how they did custom lighting. I really feel for those guys, it took me ages to get to where I am and it is non-trivial. But hey, maybe I'm an idiot so it just takes me a long time.
Anyway this is me (light and easy of course) in my "voxel lighting". It makes me think I am on the right track. I do effectively custom lighting and it took much trial and error to get it "right" and understand things. But I feel it is probably the thing that add atmosphere and worth it totally. It is hard in the sense, but it takes a lot of 10 steps forward and 9 steps back. But on the other hand easy in the sense I was able to do it using fairly basic techniques. It works wonderfully with the forward render. I see my lighting model on par with all the effort with supporting arbitrary gravity.
I'm also a big fan on playing to the strengths of your engine and not fighting it. I think I'm seeing the benefits in this while integrating the VR movement as per robo recall. The biggest issue I have is this non z axis gravity which effectively means I have to turn blue prints into C++.
the video I watched is here...