[uTerrains] Ultimate Terrains - Voxel terrain engine

Ok awesome, thank you, i will do, ill just keep fiddling and learning until then

Will the next update still be compatible with 2017.3?

Or will it be making the jump to 2018

 

That's what I forgot to mention, it won't be compatible with Unity 2017 anymore, but will require at least Unity 2018.1.8.
This is needed to get stable .NET 4.x plus a few other things.

I hope this is not a big deal for you.

 

I have a question for Ultimate Terrains users: do you need the bake feature?

It appears that it is approximately as fast to generate terrains as it is to load them by deserialization. Also, baked terrains are limited in size because they use a big amount of memory.
Personally, I can't see any reason to keep this feature... but maybe you do!

Please, let me know!

Edit: with the upcoming update, loading baked terrain is actually slower than generating them.

 

I'm not currently using the baking stuff

And I'm excited for the update, ill try and update my project to 2018.1.8 now so I'm ready

 

This asset looks really interesting to me, and the blending of smooth contours and sharp features is exactly what I'm looking for. However, I'm curious about performance. I tried out the demos and found them to be a bit "hitchy" with stutters every so often when moving around (with LOD popping in I assume) or with editing terrain.

For the project I want to do, I would love to be able to use a tool like this to "sculpt" terrain but not need the real-time aspects of it. But I'd really like to have it be able to run on hardware like the Nintendo Switch, which is closer to "mobile" performance. I'm willing to make sacrifices to get it performing like that in terms of map size, turning off map editing, and so on, but I wanted to get your thoughts on that.

Thanks!

 

@AmandineEntertainment I don't think i want to update anymore, everything's starting to slowly come together and i worry updating unity versions risks breaking that

But id really like the smooth operation script from the next update

By any chance could you post a copy of the script here for people who don't want to update to use?

 

Hello this asset looks really amazing and i'm very close to purchasing it. I just have a few questions.

- Can the CTS terrain shader be used?

- The description says heightmaps can be used. Does this mean I could use an application like World Creator to build my terrain and use uTerrains to load in the heightmap?

- Can I use something like Vegetation Studio for my grass and tress? (note: the new version contain support for mesh terrains, so I wanted to know if uTerrains could be considered a mesh)

- I want to create a fully handcrafted terrain (non procedural) that’s the size of 40km x 40km. This would be split into hundreds of smaller sections and loaded when needed. Is this possible.

- If the answer above is "yes".. could those smaller pieces be turned into prefabs so that I could use it with a system like World Streamer?

 

Sorry the script is not ready yet (even in the new v3.0) but when it will, I will publish it on uterrains.com
That's too bad you can't update to Unity 2018.1.8 (or further).. there are many improvements over there!

Not sure of that. If they provide a triplanar shader that supports vertex color control, then yes.

Yes, you could but to be honest the heightmap importer needs to be improved to smooth terrain (by interpolation) between heightmap pixels.

Ultimate Terrains are made out of meshes, but they use several meshes (for chunks). I'm not sure at all vegetation studio would work properly.

You could split your heightmap into multiple pieces (let's say, 40x40 heightmaps) and load them on the fly. You would need to implement your own 2D Generator to do that however. Maybe I'll add something like that in the next update.

You can make prefabs out of uTerrains (which will naturally include all their settings including generation params). However, I can't guaranty this will work with World Streamer. But when you'll see what you can do with uTerrains, maybe you'll figure out that you don't need it as uTerrains can handle infinite terrains.

 

Please, check out the new demo made with the brand new v3.0! https://uterrains.com/demo/

 

Hi there,

Just a word to say Ultimate Terrains 3.0 is out!

I'm really happy with this update as it bring very big improvements.

First, the overall performance has been improved. Now, Ultimate Terrains makes an optimal use of CPU cores. Pooling has been improved to avoid making GC alloc at runtime (to avoid garbage collections).

Then, you have now the possibility to perform operations on the terrain asynchronously. This becomes the prefered way to perform operation as it prevents lags.

Finally, internally, it's a big refactoring making the code cleaner and more maintainable in the future. This is key for next updates of the product.

And of course, this update also comes with several bug fixes.


For those coming from v2.x, please read the Upgrade Guide.
In particular, Ultimate Terrains now requires Unity 2018.1.8 or more recent.

 

3.0 news sounds great! I think I'm going to buy it this weekend. Also thanks for the answers earlier!

I have one other question. Let's say I create my world in normal Unity terrain. However I want mountains and large hills to be voxels (for caves and dungeons).

Could I make a simple mountain in uTerrains and bake it (turn it into a mesh)... then add it where ever I want on my normal Unity terrain? If so, then this asset would be good enough for my needs.

 

hello

where would i find the code for JSON serialization and the save/load functions?

i am attempting to network uTerrains and you must have had something in mind for this when you wrote the code.
would you mind giving a quick overview of your thinking or some clues how this was intended to work please?

thanks

 

It would be better in terms of performance to have only one uTerrain in your scene that generates all the mountains you want. Nothing prevents you from adding standard terrains to the scene too, so you could use standard terrains between mountains. However, placing mountains where you want might be harder than what you think, because you will have to implement your own generator for that. I am just about to write documentation about writing it's own generators and so on. It's a matter of hours/days. But even with the doc, it's not that simple to implement it and requires a bit of trial-and-error method.

Same thing, I'm about to document it (probably today). You can use these functions from the UltimateOperationsManager:

Code (CSharp):

1.  
2. string SerializeOperation(IOperation operation)
3. IOperation DeserializeOperation(string json)
4. string SerializeOperations(List<IOperation> operationList)
5. List<IOperation> DeserializeOperations(string json)
6.  

The UltimateOperationsManager can be retrieved thanks to:

Code (CSharp):

1. terrain.OperationsManager

And of course you can retrieve the terrain itself like any other component in the scene, with GetComponent or
GameObject.Find or whatever.

 

thank you i got networking working

another question if i may,
the trees and grass, if i add other trees or grass in the inspector only 1 of them is generated
do i need to change anything to have more than 1 tree or grass type on the same biome?

 

Be careful with the priority of your different grass materials / tree objects. Let me know if it was because of this.

 

What can you say about memory leaks? I played in the demo for about half an hour, infinitely digging a mine. Initially, the demo occupied about 1 300 mb, by the end of the session it weighed 2 069 mb. Is it a memory leak?

How are saved the changes at the level in the demo ? The demo does not seem to create any files in user Documents. Maybe because of this, the consumption of RAM is constantly growing?

Also I have 2 unpleasant crash. During landscape editing (the second time I added ground at a level with mountains), the monitor shows a warning that there is no video signal, then a black screen. After calling the task manager via Alt + Ctrl + Del, I was able to see that uTerrainsDemo.exe still works. When i switch to the game, nothing new appears on the screen, but the mouse twitches in the middle of the screen and can not move from there. Unfortunately, I had to forcefully terminate the application. These bugs happened twice, when I played for a long time and did a lot of editing. There are no error messages. It seems that the game continues to work, but the image disappears.
My computer: Windows 7 64 bit, Intel Pentium IV 3 GHz, 4 GB RAM, GeForce GT 730, NVidia Driver 390.77

 

@Lesnikus5 thanks for your feedback. There should not be memory leak but I will test again to be sure. In the demo operations are not saved to disk at all and are all in memory so yes from this perspective it is a memory leak. Will improve this in the future. However operations take very little amount of memory so I doubt this took 800mb of memory even if you dug very deep.

About crashes, it's hard to tell what was the cause of it without more information. Never got this on my side. Do you know what was the amount of memory taken by the demo just before it crashed?

 

Yes, 2 069 mb.

 

I get a weird stutter every few seconds (increases the more edits made), could this be related? i thought it was just my pc?

 

This is not normal. It generally happens because of garbage collector, but everything is pooled in ultimate terrains for this very reason. Does this happen when you don't edit the terrain too? Also, about the crash, did you find any log files? (I sent you a Pm about that)

 

yes that was it thanks.

about the stutter mentioned by joferjoe
i have tried uterrains with 2 different assets for player controllers etc
one of them works flawlessly, another i tried recently i see this stutter.

 

Can you tell me which player controller has the issue so I can check it on my side? Also, what is your hardware configuration?

 

sure
with the imminent demise of UNET i was testing other solutions and tried snet
around every 3 seconds the game paused very briefly. it was a clean install of snet and uterrain.
i added the player spawner to the mountains example scene.

the machine i tested on is i7 gtx 750 ti

 

just tried it with UFPS multiplayer and it runs fine on a much lower spec computer.
i5 gt720m

perhaps there is some kind of clash with certain other assets?
maybe joferjoe could say if there were other assets involved

 

Thanks for your replies. Then it's probably because of SNET. If it does GC allocations for example, the garbage collector is called and it can be the cause of lags. But I will look at this just to be sure.

 

This is just uterrain in unity with no other assets on the mountain demo scene
By default they're about 15 seconds apart
But the more i edit the terrain the more frequent they seem to become
At first i thought it might be the lods but i tried without any lods for a while and it still happened with lots of edits
https://my.mixtape.moe/liogpt.webm

 

Thanks for the video. Yes, I know this happens inside the editor but it should not happen on final build.

And I can explain why: Ultimate Terrains takes a lot of memory (which is normal for a voxel terrain), especially when cache is enabled (you can disable cache by setting Max Level With Cache to 0 in global settings) and Unity's editor does a lot of garbage that forces the garbage collector to run often. As a lot of memory has been allocated by uTerrains in the first place, the collection takes a lot of time.

Unfortunately, except reducing memory usage (which will reduce the issue but won't fix it) there is not much I can do. I think this issue will be completely fixed as soon as Unity will have upgraded their garbage collector, which is fortunately on the roadmap: https://unity3d.com/unity/roadmap

 

Hello! I am just looking at voxel engines for a open world fps game, Uterrain I here good stuff about!

I have a few questions if its alright.

Any texture can be used on the terrain right? and any gameobjects can act like trees and grass/objects?

I want to try using megascans, they have grass/trees/etc. didnt know if it could work with UT.

Sorry if those are dumb questions just curious!

Thank you!!

 

Yes any texture can be used, and you can even use custom materials like RTP3 triplanar vertex control shader for example.
Any gameobject can be used for trees. Details objects are GPU instanced meaning they can have only a mesh and a material (no colliders or any other thing).
Grass is procedurally generated, so it cannot be made out of gameobjects.

Haven't tried megascan but I'm curious about that too. That would be great if it worked with uTerrains out of the box! But I can't promise it will.

 

Thank you! Also can there be multiple biomes in a infinite procedurally generated terrain? Like if you were to keep walking snow would appear, more mountains. Or walk to a lake, come into a forest.

Thank you!

 

Definitely. The biome system is made for that. Just keep in mind you will have to learn how to configure biomes to do what you want and maybe also implement your own combiner if you want to get exactly what you have in mind. Don't be frustrated if you don't know how to configure everything in the first place. Ultimate Terrains is a big asset and needs time to master, but once you understand how it works, you'll have absolutely full control over terrain generation

You can also get a look at the documentation on www.uterrains.com

 

Thank you!! Yeah I scrolled through the website and was thinking to myself why I asked some questions in the first place

One more question. Does it work with multiplayer? Like the digging I know can work. But I’m thinking like the position of the player on the infinite terrain? Can they be in the same biome if they are near each other? I saw on the website a save/persistent system. Would that work with this concept as well?

Don’t worry this is last question and I’m good! Thank you for your time!!

 

Sure it can work in a multiplayer context. The terrain itself will be strictly identical among players like a standard Unity terrain. You don't have to bother with this. The only things you actually need to send to other players are operations (dig, etc.) but like you said it's easy. You'll just need to serialize them (there are methods for that), send them through the network and deserialize them to apply them for all players.

No problem, you can ask as many questions as you want.

 

Awesome!! Thank you!! That’s all the questions I have for now

Thank you again!!

 

i have been doing some profiling on snet and another asset with uterrains.
it turns out garbage collection is causing the stutters when the memory used per frame goes above around 2mb.

is there any way to combat/limit this?
thanks

 

How do you propose we make realtime in-game player terrain modifications persistent with uTerrains?

From the docs:

"The simplest way to save all operations made on a terrain is to call the Persist method of UltimateOperationsManager. This will save all modifications to the appropriate folder (which is Assets/StreamingAssets/uTerrainsData/<terrain-data-path>/operations).
Note this method is not suitable for realtime saving as it may be too long."

When I first used this asset i thought it was strange it wasn't saving, poked around in the code and found the Persist method and added it to the player script / method for digging. What I noticed though is that after every each Persist() it rebuilds all changes back to all the operation files after purging them, which obviously won't even come close to scaling.

Are you saying the voxel terrain is entirely editable at realtime but no suitable runtime method for disk writing the operations is available? If so doesn't that seem to limit the Asset significantly? And obviously the option for the player to save terrain manually is out of the question.

I hope i'm missing something, I've spent hours trying to understand your approach, as this seems strange to me as I'd consider this an essential core feature of such a system.

Not without significant effort getting up in the guts of your architecture I can probably write this myself, and have previously done so for other similar terrain projects, but in the spirit of using your particular engine, not the least of which is due to it being amazing and fast, I'd like to avoid straying too far from your lovely code

 

3.0 is great thanks for that. I think one of the best cases for unity's new job system is voxel engines and hopefully you'll put that into good use

 

@BradZoob Believe it or not, it's actually a missing feature and I realized that as I was writing the documentation you quoted! Everything has been though to make it simple to save modifications on the fly, per regions, except I didn't implement the method do to so automatically... It is really something I forgot to do.
It will be done in the next update

@Ashkan_gc Thanks! I've looked at the job system and I'm not sure it will be suitable for uTerrains, at least not in the short term because it is very limiting compared to normal threads that uTerrains uses. Also, I'm not sure there would be much gain. We'll see

 

Just a little favor I would like to ask you guys: if you bought Ultimate Terrains and if you are happy with it, I would be extremely thankful if you'd write a review on the asset store (or just rate it). This is very important to me, and probably the best thing you can do to support the product

Some fresh news: the next update (v3.1) is already on the rails. It will fix a bug where vertices are sometimes outside of their intended area and create ugly artifacts, as well as an issue in LOD transitions where small holes could appear sometimes. It will also bring all methods needed to save operations at runtime, on the fly.
I'm also still working on the documentation.

 

Would it be difficult to add building destruction to this system?

 

What do you mean by "building destruction"? Buildings made out of 3d models or buildings made out of voxels?

 

Will the next update require a new unity version?

I hate updating unity mid project, but i love the sound of the new update

Also will there be smoothing operations added in the next patch?

 

No, it won't. Maybe I will even make it compatible again with older Unity versions because I realize that many people don't have installed Unity 2018. However you will still get performance improvements by using Unity 2018.

I have some bad news about this. I will do it but probably not in the next update because it is actually quite complicated. Smoothing requires to know the value of neighbors voxels and currently operations are not designed this way. I need to think about it from another perspective and it will probably be a lot more work than just creating an operation

 

Yeah, it's crazy how much performance you get in 2018. I finally upgraded a big project from 2017.1 to 2018.2, which i was previously hesitant to do, but the gain was crazy for my particular system which was all about procedural noise, building/updating chunks etc. And the upgrade was pretty smooth, a few modifications were required here and there, but all up about an hours work, way less than I was preparing myself to do.

 

Hello!
I decided to take the plunge and purchase uTerrains. I've been playing with it all afternoon!

A few things I'd like to ask after having played with it a little:

1. Is there any way to use the same terrain in multiple scenes? For instance, maybe the game has multiple seasons and I want to use the same terrain but across multiple scenes so that later scenes have different settings on the same terrain data.

2. Is there a way to export the terrain to a mesh/series of meshes? The reason I ask is for a couple reasons:

• (a) I'd love to use the terrain editor to make a custom terrain but maybe not have it be "editable" in realtime -- seems to me it could maybe perform faster if it wasn't generating in realtime? I could be wrong about that.
• (b) I was thinking I might like to have a certain area that is explorable/editable by the user, outside of which is a boundary. The boundary I could use static meshes and perhaps even use them as "imposters" at a distance, thus improving performance while keeping the same great viewing distance and same type of texturing, etc. I'd know the player could never reach these areas so I could also perhaps do some other optimization on these meshes.

3. What method would you recommend using to see which type of voxel a character is standing on?

Thanks so much! I'm really enjoying playing with the tools.

 

@BradZoob yes 2018 is really a big improvement in performance, and even more when you select .NET 4 as Scripting Runtime Version. I can't wait for the future garbage collector. I don't know if it will come with 2018.x but it is really a pain point today!

Thanks for buying!

You can simply make a prefab out of the uTerrain object. It should be enough

No, there is no way to do that out of the box,and this is for a good reason: without the LOD system (which obviously can only be dynamic), there would be way too many vertices to draw in your scene. There used to be a baking module that has been removed because it was actually slower to load baked data from disk than to generate the terrain normally.
Also, the terrain won't be faster if it is not editable. However, if it's not editable then you probably don't need cache, so you can set "max level with cache" to 0 in global settings which will save a lot of memory.

You can still do that, it may be a good idea for your needs, but you'll have to pick up chunk meshes by hand and save them somewhere. I guess the Editor Tool might help you doing that.

Get a look at the PlayerBuilder.cs script. It gets voxel info when you right-click:

Code (CSharp):

1.  
2.                 if (Input.GetMouseButtonDown(1)) {
3.                     Voxel targetVoxel = Terrain.GetVoxelAt(Terrain.Converter.UnityToVoxelPositionFloor(wpos));
4.                     if (targetVoxel != null) {
5.                         targetVoxelText.text = targetVoxel.ToString();
6.                     } else {
7.                         targetVoxelText.text = "";
8.                     }
9.                 }
10.  

The method you are looking for is Terrain.GetVoxelAt() which takes a voxel position as parameter.
To get the voxel below the player, just perform a raycast to find the ground and the get the voxel at this position.

 

Thanks for the responses, @Amandin-E !
One other thing I came across while playing with it -- it seems like only the first 2D generator/3D generator in each the biome lists are used, so why can you add multiple ones? It feels like I'm missing something there, and the documentation at this point says to read up on it .... nowhere since the link doesn't exist yet XD

 

Which 3d/2d generators are used depends on the final combiner. It's true that most final combiners actually use only one or two generators, but you may want to implement your own combiner that would need more. UTerrains gives you the possibility to do that simply. It's all about freedom to customize

I'm going to add more combiners soon so you'll be able to do more out of the box.

 

Making good progress understanding the system... the variables are REALLY important, you can get hitching pretty easily if you put one too high or too low.

I was wondering what I can do to reduce instances of this:


I'm using the same Global Settings as in the example files and the 2D Master script settings are the same as the examples too. I also get occasional jagged polygons in weird places/gap in the terrain... I'm using the MultipleHeightmapsCombinerSerializable just for reference, with the values set to 0, 1, 2 respectively. Anyway, thanks for the continued support and development! I'm hoping the documentation is updated soon as well