Hi!
I want to create my first map and I don’t know how to start. Searching on the internet I’ve found the landscapes, but I don’t know if every level has to use landscapes.
When should landscapes be used?
By the way, I want to create a sea port. In this case, do I need to use landscapes?
Thank you.
What familiarity level are you at with other software? Have you worked in anything before unreal?
No. It’s the first time I work with a game engine.
Never.
Its much better if you just work in your DCC of choiche and build whatever you want to import as an appropriately sectioned mesh.
Also what? Did you steal your Dad’s acocunt or something? XD
No.
Maybe the answer is in this other answer: Should i use landscape or can i just use "blocks" - #4 by jwatte
They cover good points, however there really is no reason that anyone making a game should ever use the landscapes except for initial prototyping - and even then, those who choose to must be extremely aware that before shipping they likely have to replace everything with meshes in order to hit the performance requirements anyone would expect.
So, when should landscapes be used?
When prototyping, maybe.
And you should make sure not to use functions that lock you into them like the landscape grass for instance.
Or when not making a videogame.
But can’t the landscape system be optimized?
NO.
The only way to fix the increased costs it brings to your scene is to replace it with meshes.
So, even when prototyping, it’s likely better to just use something like Houdini.
If you make a very large terrain, the landscape will be a smaller install/disk space than the equivalent meshes. So, for large area games that are downloaded, the performance cost at runtime may be OK as a trade-off for a smaller install/download.
Agree to disagree. The size of the mesh and the amount of tris it has (plus LODs) are what drives the size.
Your avarage landscape level in .26 was around 2GB for 8k res.
How many meshes can you make before you hit that threshold?
How many games have 8km^2 maps? To put the size into perspective:
Are they nanite, like almost everyone is moving toward these days? If so, I can fill that pretty quickly. If you’re making an entire landscape out of meshes, I hope you’ve got a good PCG scattering algorithm. Though you could use a low detail landscape to place the meshes, turn them into a PCG stamp and set the landscape to editor only, if you wanted to.
No need for pcg. The procedural tools existed way before that even in .18 i believe.
They have just never been finished like everything else within the engine, but they do work.
Also 8km^2 is pretty small.
Largest I have is 244km^2 and that’s still small but because it was landscape based it is about as many levels as you can cram in before the drawcalls get to be too much when looking at it all.
Not that it matters, but rather sure you have witcher 3 wrong. They were supposedly using tile sizes of around 1km^2 - you really wouldn’t need to partition things for the size you listed.
Same with many of the other games you listed.
Anyway you want to look at the other ones, and probably the ones built on the engine?
Pugb, fortnite, etc.
Maybe even just the Kite demo.
Re Nanite, don’t know.
Usually you do not gain performance from using it.
Looking back on it, I shoudn’t have put the ^2 at the ends of them, since it looks like it’s attached to the kilometer unit and should have just said “12km by 12km” or “12x12km” or "(12km)^2.
So the list in terms of square kilometers would be as follows:
Yes, but they are still PCG, just not the plugin version. PCG stands for Procedural Content Generation, which is a generic term for content that gets generated procedurally.
And minecraft can technically scale up to the surface area of Neptune or something around there. It doesn’t mean it’s interesting.
Then you’re using it wrong.
Back on track: Stop spreading misinformation about landscapes being unusable. Plus, you completely left out all the optimizations pertaining to collisions and LODs.
The only one off track and spreading misinformation here is you.
PCg is not a “generic term”. Never was. It was specifically added to the engine and became popular as the related PCG components came about.
Prior to that it was and is still called procedural foliage.
Then you spread misinformation on game maps sizes. Double down on it too.
Then you blabber off about fantomatic “optimizations” that don’t exists and have never existed in engine in any way, shape, or form that actually does anything for anyone.
And you try and preach about spreading misinformation?
On a topic on which I literally wrote books, to boot.
Do us a favor and get off this forum instead of blabbering on.
If you’re making content procedurally, then that’s procedural content generation… Kind of like how apps like Photoshop, Zbrush, Substance Painter, any digital app used for content creation, etc, are sometimes called DCC apps, short for digital content creation.
There are a staggeringly large number of sources for almost all of the games I’ve listed… I’m not going to spam this thread with a million sources for each one. The sizes I listed are the rough square sizes. Many of them have non-square shapes, but when you take the total area people have calculated and then take the square root of that, you get the rough dimensions of a square map that would have the same area. For instance, you could have a 2 x 10km = 20 sq km map, which would make a 4.5 x 4.5km map if it were a square. For complex shapes or calculating actual playable area, you can use things like splines and flood fill algorithms to closely approximate; so long as you know the scale of the level.
Lots of people write books, most aren’t noteworthy.
But I’m not going to entertain your argumentative forum trolling any longer. You’re wrong, it’s that simple. Landscapes are still plenty viable for actual games…
They are not.
Its that simple. There are around 1000 examples and explanations. Do your research.
I have. I’m looking at my 4km map that has a bunch of layers of single and triplanar mapped texturing, distance blending macro variation logic, landscape set to nanite with tessellation enabled, RVT color/height/displacement, completely covered in dense nanite foliage and I float between 80-100+ fps on epic preset depending on where I’m at. If I disable tessellation, the fps is around 10-20 fps higher on average.
This has to be one of the most odd and niche “small hill to protest on” that I think I’ve seen in quite some time.
You have 1 landscape running. What you have is at best a demo. Not a videogame.
Quit thinking you know stuff. You don’t.
Quit thinking you should derail some other noob’s project by sharing that “it works”. It doesn’t.
You are so green you list FPS instead of MS without any context what so ever. We’d be lucky if the pointless stats you report are even from the packaged launched build.
Correct, it is an interactive tech demo. There are a bunch of other benchmarking loads placed in the world, like roaming fauna, heterogenous volumes, particles, interactive day/night cycle with weather, UI, multiple movement modes, post processing effects, a fairly complex water shader, etc.
FPS is a completely viable stat to use, if you have two points of reference. 100->120 fps is the same as saying 10ms->8.33ms, since 1 / 10ms = 100fps and 1 / 8.33ms = 120fps. The times when it’s a bad stat to use is when someone just says “I enabled X and I lost 60fps” without giving the other half of the context like starting at 250fps and dropping to 190fps vs starting at 100fps and dropping to 40fps. Massive difference between the two. I used it, in this case, because it seemed simple enough for you to understand.
Packaged for shipping.
Hey genious, The context would be the output size and hardware, not whatever other Bs you feel like writing down about Ms being the same as FPS - which they are not.
And your “tech demo” is wrong. At a bare minimum you would need 4 concurrent landscapes present and loaded to even consider it a “tech demo”.
You are also way off topic and infringing forum rules.
This isnt a “look at me, I’m so cool I can make the engine run even though anyone else can too” topic…
Its a specific question about when if ever the landscape system should be used.
Sure, its also been heavily addressed, so no one probably cares whatever else you or I add here, but make your own topic if you want to share pointless things again.
And you’re spreading misinformation, claiming they should “never be used,” just because you couldn’t manage it correctly yourself. Your limited results are not indicative of the overall state of their functionality and if anything, I’m guessing you didn’t take the proper steps to actually optimize your world and didn’t even take the most basic of first steps of optimization like using world partition(correctly).
Meanwhile, I’m giving my own current personal results and your claims just simply aren’t holding up. And it’s not just this current project I’m working on, it’s been roughly a dozen now, some of which were actual prototype games with real gameplay(lots of placeholder assets, but almost always of similar poly counts and material complexities to what the real assets would be). Not to mention all of the UE4/5 games that use landscapes just fine.
I know cognitive dissonance can be hard to deal with, but there are plenty of coping mechanisms you can research and learn, that don’t involve forum trolling. Maybe even try asking ChatGPT if you’re having issues understanding it all? It’s pretty good at summarizing things and you can chat back and forth with it until it finally clicks.
But to accuse me of being off topic for correcting your blatent misinformation, just because my own personal results don’t align with your claim… Don’t even get me started…