it’s still an open question without a proper workflow, will depend on how Nanite evolves as well as how Landscape evolves (perhaps one day Landscape becomes Nanitable, since the ultimate goal is to Nanite everything)
for now using WPO instead of displacement can do the trick both for Landscape and having to use non-Nanite meshes, however the mesh density needs to be quite high
Studios should really share/contribute code(quality code of course) to Unreal, it will benefit all users not just Epic and anyway, game studios sell game not code. I hope we start to see similar happenings in future. Maybe it has happened, i just do not know.
I’ve been thinking about that trailer ever since I first saw it.
“How can I make snow like this?”
Deep snow is probably not that hard but the tiny snow? It might just be an effect and not actually snow falling like you’d imagine. Imagine adding pre-rendered fluid dynamics but with snow. Snow dynamics? It might just be a trick and not an actual function. I could be really wrong though.
It’s early access
Correct, not a finished feature
Wait until it’s finished
Epic isn’t interested in answering a barrage of questions over a work in progress, not yet meant for production engine for which they themselves probably don’t have all the answers yet anyway.
I know it’s not the answers everyone here wants, but those are my guesses as to what they would be anyway.
You guys lol, have you thought about the trailer just being faked? and what RECENT games that have come out solves these issues or is a fully next gen sandbox using unreal? only thing I can think of is cyberpunk and all of unreal issues show in that game.
a glorified tech demo, not an actual game…yet.
Ever see those Ocarina of Time demos running on Unreal engine?
I doubt it’s fake, but I’m sure it’s something they built themselves. Gears5 used UE4 which had functioning tessellation, and yet they still chose to build their own tessellation solution to handle snow.
Disagree, they are working for me: https://imgur.com/gallery/Ws9Y2Na Seem to be quite performant COMPARED to what landscapes/tessellation used to offer. Overall, sure I’d wish they perform better, but that I can still get triple digits on a full 2k landscape like that, I’m ok with.
Once I get my render targets set up and hooked in, deformable snow ought to be a go. This is all tech that should still work in UE5, but I’ll test/post.
What I can say from my experience is that something like snow where i just futzed with the basecolor, pbr, on a 2k landscape w/1megatexel (14,1,2) RVT, it would update realtime on the whole thing bit in slightly under a second. You could see all the tiles flicker/update rapidly. For smaller things, it should work better but I have yet to test. I was trying for worst case scenario there in being zoomed out looking at the entire thing. If you are in a valley or likely only to see a subset of landscape tiles, it would seem to be ok.
You can also do stuff after the sample from the VT. That will work in real-time as it’s not something in the VT itself, but on-top/after. Generally, I find, keep the VT static, but you can trade that for upping the res/detail given you can also enable streaming.