Floating Island Assistance


I am thinking of creating a level which is a 2km x 2km floating island.

I initially stayed away from the idea due to not knowing the best way to tackle this. I am pretty sure the Landscape tool can’t be used to accomplish this due to the floating island shape. So I would assume you would have to use several static meshes to build this up.

I guess I’ll list some questions below and if you could help with any that would be great.

1.) If I use Static Mesh for the ground terrain which the character is using to traverse, whats the best way to handle accurate collision?
2.) For a 2km x 2km island how many pieces should I break it down into? (the more the better i guess for culling) To also add if there is a high up vantage point could I see the whole map or would aspects of the level unload (leads me to me next question)
3.) Should I use world composition for this? The project in question will support 4 players so I am unsure the compatibility with World Compo and Multiplayer. To add If a cave system is added I’m almost sure World Compo will be needed?

I searched for Floating Islands and UE4 and unfortunately didn’t turn up much. If anyone can check this out and help me out I’m sure it will help others in future too.


Cheeky bump, maybe I could use a landscape for the main part of the island and encompass it with static meshes to make it seem like it is a floating island that may be the bet way. If anyone has some thought please free to chime in

Are you ever going to see the bottom of the island?

If you don’t you could sculpt the landscape into a kind of roundish shape, then hide the corners underneath cloud particles and fog?

It would be nice to see the island from a distance as in a rope bridge to a much smaller static mesh floating island, or a waterfall dropping off the edge of the end and possible platforming around the side of it. It’s a shame you can’t shape/delete the landscape in a circle sort of shape. I might use your idea but mask the edges with a few static meshes to simulate the look of a floating island. As far as I can see that might be the best way unless anyone knows anything different.

Thanks for responding

I am working on a similair idea, I’m gonna attempt to use smaller landscapes to flesh out the major shape of the island and use static meshes for the edges. I had an idea I have not yet experimented with to use a one sided material for the bottom of the whole and since it would mostly be visible from a distance the mesh doesn’t have to be to complex for me.

I would make an rough island mesh in Blender and after dazillion times of reimport and the scale is right, add “blokish” details.
Then walk through map, check viewpoints, place blocks, where you want to limit view, remeshing again later.

The rough mesh could act as placeholder, where you place the smaller parts of mesh, to “overwrite rough one”
When you need a lake somewhere with 200x867.30m, mesh one import, place, reshape.
When you want to go the realistic way, use some rocks from the provided content and scale them up to something like an island.
Use smaller scaled rocks/ to get variation.
I did that some time ago and it was nice, because every sector was editable, but i am not sure about performance.
I took only the lavastuff and some rocks from elemental demo, to sketch out a map.
Later i meshed fresh lavameshes in the shape i needed. Was a floating island too.
When you get this “big static actor” lighting msg, reduce size, till ue4 is happy and work with that(when you need dynamic stuff).

That’s not an bad Idea… maybe make a rough shape and any 3d modeling app, Very low poly then divide the mesh in to sections using the current edges freeze the outside edges and the subdivide and go smaller… hmm :slight_smile: Lots to think about :slight_smile:

You still can use landscape actor as it doesn’t have to be a perfect square. After you add landscape you can delete sections of it to get a rough shape and blockout ends with pieces of static mesh. Doing it completely out of pieces of static mesh is definitely an option but for 2x2km size that’s a lot of work.

You can actually model a basic shap in a 3d app and use mudbox and world machine to make stamps for mountains and other features. Combine the stamps to create a unique terrain. Been toying with that tonight and will post some images when I have something that looks decent :slight_smile:

Yes with those external tools you could “easy” create cool looking scenery conent, but opting is another thing for gaming/performance.
You suggest learning three external programs, to get a basic shaping solved?
From my point of view that’s to complicated to get started, all those learning curves external programmstuff and then importing testing, then performance, could to easy break motivation for me.
Why reinventing the wheel, when you in the end see no difference?

Geez, I didn’t expect this to get any responses back, I appreciate it guys.

Yeah so what we’ve done is created the Island shape in Blender and exported that out to UE4 then added a landscape and essentially sculpted the landscape to match the shape of the Static Mesh. This has worked quite nicely, then we trim the squares that aren’t used from the terrain and the idea is to use 2-3 different Static mesh rocks (with different scale/rotations) to make the sides of the “floating island”.

I have to say my knowledge on performance is not great so it’s hard to say which path to go down. The reason we didn’t go all the way static mesh is because the UCX collision just isn’t accurate enough when traversing the landscape on foot. (You would float on certain parts and collision wouldn’t exist in some)

The perks of the Static Mesh route is as you said is pretty modular and allows a great deal of customization however I guess I could add some Static Meshes for some customization. If we get something pretty ill most likely post a screenshot if it interests you.

I agree with LuftBauch about using Mudbox/WM its kinda pointless for this task. Also i find when working with medium size maps like 2km (which we may shrink to about 1.5km) the features like erosion etc create relatively large ditches which makes traversing uncomfortable. Its visually appeasing but not practical when you want to traverse.

This is also dependant on what kind of brush you make ofcourse. But I agree it might be better to hand craft terrain to suit certain needs. And I did not account for the learning curves because I already know them, all comes down to preference in the end I guess :slight_smile:

:slight_smile: Nothing i write is meant negative or as an insult, only my 50rupies.

When you know the workflow, because of experience in those external programs it’s nice, but hard to get to the point for a beginner.
For me personal i stopped nearly any work with external programs, aside Blender, UE4, Audacity, Krita, Inkscape, Papagayo and i bet some more.
That’s still a bunch of learningdata for my small singlecelled brain.

I made that decision with full force, where i was at a point i had to buy substance for realistic stuff(money was not the problem), because all want pbr these days, everywhere.
But then substance will quick get a mainprogram, no longer some sort of helper for speedup tasks.
Much time is needed to use full potential of that, together with normalmaps and stuff, so much to learn.
So much to learn that it gets quick a focus in freetime and other things have to stay back more then.
That counts for any external software, updates could cause bad headache and relearning stuff, or it’s causing errors, result in changing workflow.

For me nothing has to be perfect, because nothing is.
I am single dev and 95% of time happy with that.
External program usage/learning stuff would stretch projecttime, even when it speed up workflow in the end.
All these tools are cool for sure, at least at some niche.

Stuff i ask me often: “But how much is my time worth to invest in that - is there no solution in range (could be limited one), with my basic tools?”
“When i find a workaround in tools i still know a bit, i gain experience in my mainprogram”.

I learned so much last years and in the end for what?
I am no pro in a company/team and specialist for landscaping, why i should start using that?
I like to be a creative rapid prototyper, even when i could invest only some hours into my project daily(1-4hours).
Whole day running around with my sketchbook(something like a crypted partial GDD), where i plan everything ahead.
Thats the first thing i see, before i rest my eyes and the last at the end of a day.
Sometime i wake up in the night with an idea/solution and write it down with a smile.

Sry for blahblah, but this gamedevway makes me personal very happy, some sort of progress motivated happiness in a neverending loop.
Feels good. Al those failed superrealistic projects i joined, last ten years, died halfway up.
All that time feeled like trashed, but now all that makes sense for me.

No offense taken or ment luftbauch :slight_smile: I’m also a single developer in my spare time even. Nothing note worthy that I have release, pure hobby atm with plans for releasing something in… let’s say 2019… 20… 21… and so on :stuck_out_tongue: Just rambling thought as you do trying to contribute ideas and fetch info :slight_smile:


So I did floating islands a while back. The way I handled the collision was generally to use a lower-poly version of the model and changed the COLLISION COMPLEXITY to USE COMPLEX COLLISION AS SIMPLE. In general this worked pretty well but I not sure what impact it would have on performance. Later I just used a flat plane as the top as there was no real collision required at the bottom. I suggest using Blender’s SHRINK WRAP modifier for your lower poly collision box.

Have a look at the screenshots. Very rough but it works.