Is there a voxel-based terrain implementation that's any good?

So I tried TerraVol but it wound up being broken and not working in 4.6 or 5.0. Waste of money. There is also Ruaumoko, MoleEngine, and Voxeland. None of them have been updated in about a year, and their devs all seem to be awol. Are any of them still worth the money?

I was looking for a terrain implementation that could represent steep cliffs well, and a voxel terrain plugin seemed the perfect way to do it, but now I'm not sure. Are there any other terrain utilities that are worth a look that can handle vertical faces? Preferably that can import a hightmap too, but as long as they're open/extensible I could code that myself.

EDIT: The other option is, of course, to code it myself, but I unfortunately can't spend the amount of time necessary to build a voxel terrain builder and all the tools necessary for working with it.

 

Well you don't have to use a Voxel engine to get steep cliffs, or caves, or anything else.

If you know how to 3d model, you can 3d model your terrain and make it look exactly the way you want, then get some Vertex Color/Vertex Paint Shaders for Unity that way you can paint your 3d model terrain like you do a Unity Terrain, but that's pretty much the only solid info I can give you. I don't know much about Voxels.

 

Thanks N1. I'm definitely aware of 3D modeling the cliffs as an alternative, but I'm trying to avoid that if I can. We're doing an expansive amount of terrain with primarily cliffs for elevation changes, so that would be a lot of 3D modeling work. Another issue would be that our 3D-modeled terrain would be immutable inside of Unity, which would clog up the pipeline considerably.

 

Have you seen this one, looks totally worth the money it costs:
http://forum.unity3d.com/threads/te...ral-smooth-terrain.174595/page-8#post-1999274

 

I completely understand what you mean, just wanted to throw the idea out there, because 3ds max and stuff has Mesh Deformation tools, works kinda like Z-Brush and Mudbox, you skulpt it, so give it like 10000 polygons, click and drag to make it look nice, then optimize the polygons. In like 5 minutes done LOL.

But it was just another alternative.

 

7 days to die has a good formula. They even have smoothly connecting voxels now. Not sure if the implementation details would be public, but the info for what they started with could be somewhere.

 

Actually, I may have been wrong about the devs all being inactive. It appears Voxeland's dev is still active and working on Unity 5 support, so I was wrong about that. Anyone have any experience with that?

 

Surprised no one has mentioned Cubiquity yet. It has a free fully-functional non-commercial license if you want to try it out.

 

I was looking at this one: http://uterrains.com/demo/

The demo seems good.

 

Can anyone (with more experience with voxel terrains, I guess) elaborate on the differences between Ruaumoko, Voxeland, MoleEngine, and TerraVol? Other than their reputation for support for the developer, they all seem to by very similar. I'm not sure how to make the decision, and can't afford to buy all 4 to compare/contrast.

Ruaumoko - https://www.assetstore.unity3d.com/en/#!/content/8176
Voxeland - https://www.assetstore.unity3d.com/en/#!/content/9180
MoleEngine - https://www.assetstore.unity3d.com/en/#!/content/17877
TerraVol - https://www.assetstore.unity3d.com/en/#!/content/8131

 

It appears to still be in review for the asset store, but I definitely like the list of features on the announcement thread.

http://forum.unity3d.com/threads/ultimate-terrains-brings-3d-terrains-to-a-whole-new-level.304908/

 

How does it look compared to cubiquity?

 

I haven't gotten around to evaluating Cubiquity, but I can form a rough idea from the two threads. One of the bigger differences is that Cubiquity is written as a native library so it only supports Windows, OS X, and Linux. It does not support infinite terrains and appears to be limited solely to voxel tasks (ie it doesn't handle foliage itself).

One thing I noticed from the demo video for Ultimate Terrains, that is not mentioned in the thread, is that the biome system is a visual node-based system. It isn't the fanciest node system I've seen, but it is pretty nice nonetheless.

Between that and uTerrains support for platforms beyond desktops/laptops, infinite terrains, foliage handling, integration with RTP3, and a cheaper price tag, I think uTerrains is likely going to be a better deal than Cubiquity solely for terrains. I think it is setting itself up as a complete replacement for Unity's terrain system.

Cubiquity still has some advantages as it is supposed to be capable of more than "normal" terrains. Two examples they've given were cube-style voxels (aka Minecraft-style) and more general purpose shapes (ie a planet-shaped mesh).

 

@Ryiah Thanks for the info. A voxel system is one of the few things I haven't tried to do myself yet. It's interesting that the C++ voxel system is the one that doesn't do infinite generation, since it should have the better access to memory management. Then again, minecraft has a decent system and that was written in Java. The voxels might be huge, but maybe 'voxel systems' aren't as scary as I thought.

I wonder if procedural worlds and or content will be a gaming standard in the future so that every game, no matter what it is (even telltale games), have a large degree of replay value.

 

Tomnnn if you're looking into programming your own voxels, here is a really good (seeming) blog I found on the subject: http://0fps.net/2012/07/12/smooth-voxel-terrain-part-2/

The basic algorithm most of these voxel terrains seem to be using is marching cubes, which is open licensed and linked to from this blog.

 

I'll check that out, thanks. My understanding of voxels goes as far as assigning block heights with perlin noise. I'm curious about 7 days to die's actually smooth voxels.

 

I like how 7DTD does voxels a lot. Practically Minecraft voxels but with smoothing. Doesn't work for our game unfortunately because we have a lot of landscape running in non-cardinal directions.

 

They have diamond shaped voxels for their terrain and then regular square ones when the user is placing building blocks. What kind of shape would you need for your game?

 

Hey guys,

I'm the developer of Cubiquity so keep that in mind when reading my thoughts

I think one of the most important things to realize (as is already becoming clear from this thread) is that people have vastly different ideas about what a 'voxel engine' actually is. Some people think of natural terrain with Marching Cubes or other algorithms, some think of Minecraft terrain with textured cubic blocks, and some think of Voxatron style environments where everything is built of colored cubes. Throw in the old school engines used by Commanche, Delta Force, etc (which aren't really voxel engines at all) and you have a recipe for confusion.

Next up, there is the question of what features a voxel engine should provide. Does it need procedural generation? Do the terrains have to be infinite? Should you be able to modify the terrain in the editor? Should you be able to modify them at runtime? Do you need to transform the terrain or have multiple instances? Do you want a custom pathfinding solution, or what about water simulation? There are so many factors to consider that simply calling something a voxel engine does not really help a user work out if it does what they need.

I won't say where Cubiquity falls into all this as I'm not (only!) here to sell and you can check out the link above if you're interested. But I will say that this is the reason why I make a free version of Cubiquity available for testing - without that I don't think users could have any real idea of whether it will work for them. I do believe this is a problem with the other engines as I see a lot of people asking questions about them but no one actually showing off what they have made. I think a lot of people buy these engines and then find they don't do what they need.

 

I bought uTerrains 4 days ago and I can tell it's a great solution. It is very solid and it sounds like it has a lot of potential.
I used Ruaumoko a few months ago and honestly uTerrains is already far most powerful even if it's the first release.
I haven't tried Cubiquity so I can't speak about this one but I think you won't waste your money with uTerrains anyway.

 

I've used cubiquity and I can tell from the uTerrains demo that it is more powerful (and looks better). Not gonna buy it now since I don't have €120 to spend and I have no use for it anyway, but when the time comes and Unity still won't have voxel terrain built-in, I'll probably go with this one. How does it work with Unity5 by the way? Particularly, I'm thinking of PBR.

 

I'm glad you say that. Means I probably made the right choice.
Yes it works with Unity 5 (i'm actually using Unity 5) without any problem. I haven't tried PBR though.

 

Hi!

One year later, and the discussion still seems as relevant. I've been trying to survey voxel solutions for Unity and made a List of Voxel Solutions here. Its focus is on large-scale terrains, not solutions for models or small dioramas. (The list is currently missing Mole Engine.)

If you want infinite terrain with large draw distances (and no hiccups), there's really not many options yet:

• TerrainEngine may have solid support but it costs a fortune. (around 50x that of other paid solutions.)
• Ultimate Terrains has support on paper but there was still visible seams when I evaluated it, as well as frequent hiccups (probably GB).
• Voxelland has support but has major multiple-second hiccups when generating (there's even a message saying "please wait").
• VoxelFarm most likely has solid support, but the Unity integration seems convoluted and it's not clear if it's possible to define voxel functions directly, or you have to define biomes in VoxelFarm Studio and can only use it through that system.

All in all, there is still a big gap in the market for a reasonably-priced solution with solid LOD-support (like the open-source TransVoxel algorithm) and no hiccups.

(There seems to be a C# implementation of the TransVoxel algorithm here but I can't say what the quality is like and it's not a drop-in Unity solution. Maybe it could be a starting point for something though.)

 

So, there are no raycasting voxel renderers for unity? That's unfortunate.

 

I think direct voxel rendering is interesting for the future, or for experiments today, but is it viable for general-purpose terrain rendering today? I only know of VoxelQuest that uses it (not Unity) and they use a very small screen resolution to make it possible.

 

It is probably viable, but it would require quite a bit of R&D in order to make it work in unity, especially if you want terrain to be smooth.

Last time I tinkered with non-polygonal minecraft-style renderer, the shader took quite a bit of time to compile.

Direct voxel rendering is not a new technology, it has been used quite a lot in the past. For example, in Outcast (landscapes), or in Alpha Centauri (units), or in build engine (misc objects).

 

What if you don't need the terrain to be smooth? Would that save a lot of performance?

 

"One year later, and the discussion still seems as relevant"

wow jeeze really Unity Technologies, I dunno maybe time to step up the game and like err do something about it! I've been wanting a decent affordable voxel engine since last year damnit!!! I feel like you made this post about voxel engines unaware of how things come across to general users when you work at Unity!

Anyway so I keep checking the things out in voxel terrain development.. so far I've just tested out a few free version to mess with, while it seems the only buyable version that seems reasonable all things considered (documentation etc) for smooth terrain is Voxeland. I don't think any of the others really compare much, but its not saying a lot as none of them are all that great overall. uTerrains/TerraVol and the other have some workable features, but alas they have other issues aswel.. I'm sure there perfectly good engines given the type of game, but all still leave you with a lot of things to do before even really working on an actual game.

And all of them really leave a lot to be desired on the editor tool fronts (get rekt if you think you're going to find worldedit types tools and brushes to use in building your worlds), and if you're wanting built in networking examples to work from think again, you're on your own for that aswel. Did I mention water, yeh you'll be using a flat ass shaders. Documentation, Examples, Prices etc all vary none all that good.... now good & price being subjective .. but for any indie any price can be to much really, because short of being someone who has the skills to just do stuff yourself, you're going to need to spend x time learning and trying stuff out. And in the end that asset you bought might just be completely useless for you, or you finally make a game and it actually makes back what you spent on assets for it at least or not.

I'm at that stage where building a game ontop of a voxel engine is totally within my abilities given some good example scenes, code and documentation, but building a voxel engine, making large changes, let alone building extra features is just plain out of scope or would take to long, building custom tools might be feasable, but I'd rather work on gameplay. Than making voxel terrain tools just to better work with the actual terrain. Biggest time sink would be adding multiplayer. And doing any of that still feels like reinventing the wheel.. whytf_have none of these game engines got any semi decent voxel engine out already?

Everything points to game engine makers like Unity, Unreal, Cryengine really being in the best position to really do something about this.. even from a licensing approach, and they've sat on it and left it upto third parties to offer something up... which hasn't really worked all that from what I've seen. Yes plenty of other features to work on, I agree, but yknow the first engine that really offers something good in this aspect I'm heading towards, because that's the kind of game I want to play and make. I look for those games, but the marketplace for those games is pretty small and what exists is largely dominated by only a few titles that obviously had a few key people involved to provide the tech and the rest is meh. I had the most fun making plugins and minigames within games like minecraft, 7dtd etc Look at making my own game with a world terrain like that, and options for anything really good is kinda sad.

MoleEngine got my curiosity however the current demos alone show many bugs, and its v1 development has been abandoned.. I checked in to see how MoleEngine V2 but price for early access just wasn't my cup of tea at the time of asking, without any demos to see aswel, to risky. Maybe they have something out this year.

On the free side I have looked at Cubiquity and Voxelmetric, so far the free versions that potentially offer some decent enough base for a voxel based game depending on what kind of game you have in mind.. Voxelmetric still getting some updates and its developer working on progressing things

Cubiquity, great single textured cubes that you can have whatever color you want, its fast but totally limited, no large scale support ie using thing like chunk loading/unloading... smooth terrain is also limited to 4textures, and again unless your world is tiny and your usage of smooth gen fits in your design, its not so good for actual terrain with any meaningful view distance. Only supported on desktops, ironically the engine would fit the type of simpler smaller scope mobile type game, yet no support for mobile platforms. it's biggest plus to me is "Support for multiple volumes which can exist in transform hierarchies." as hybrid approach to designing a game with traditional 3d assets, game world, and then using voxel patches of large and small sizes to fit within those 3d assets areas to provide destructible sandbox terrain is the way to go.. Something like this as a new terrain object in Unity would be great, and allow asset store devs to build tools and enhancements for something that everyone has access to via the game engine.

Voxelmetric, still early days, very much a minecraft type clone(not a bad thing if thats want you want in an engine for your game), support for multiworlds maybe coming, unet support in development.. but not stable to really go with, changing codebase.. yet the its only engine I've tried that has potentially to really be a good open source engine to go use for a cubed voxel game.

I'd love an implementation like 7dtd has which really seems to cover all the bases, 3d objects placable within the world, water blocks. It follows the minecraft chunk/region water physics approach pretty closely, it works well enough. For anyone who hasn't played the game, it isn't perfect but its voxel engine sure is one of the better ones I've seen that actually have a fun game with sandbox terrain/world built with Unity as a engine. And so far nothing for Unity on the asset store let alone github can really give any indie developer that kind of engine to build there own game on top of. Leaves a game marketplace with just meh to me as its really the only game to have smooth voxel terrain that's playable and fun.. all the others kinda blow hard and not my sort of game, and its not for a lack of developers wanting to make better sandbox games like these.. its just a lack of investment of money and skills to make it more accessible to larger number of developers to really utilize imo.

It's hard to see any game engine like unity, unreal, etc providing such a voxel feature for game developers. A flexible terrain system like that seems really complicated with to many approaches to design and performance, and making it flexible enough to be used in large number of ways.. yet we have a terrain system in Unity that frankly sucks donkey balls, doesn't even have any built in ways to make cave holes ffs.. Even unreal has a documented way of achieving it. Unity's terrain only becomes semi decent with a dozen asset store purchases by the looks of it.

Anyway I could honestly say if any game engine provided a killer voxel engine feature with something of the level of 7dtd features in which to build a game and extra tools on top of I'd jump ship, I feel like a voxel engine for today really needs a lot of resources to get done right. Many approaches have been taken so far, but the investment to developing it seems to have been bypassed by all but the small studio's and technically capable developers for either there own projects or to commercialize in marketplace where average prices don't really fund that sort of development for long. So were left with best efforts of a poor mans approach to every voxel engine on the market really. Step up Unity, Unreal, Cryengine I don't give crap out of which of you does it.... a decent voxel engine is what I wanted last year.

 

Isn't 7dtd unity? I think they got a deal with that voxel engine that you can't buy into, hah. If unity did make something that good available for free it'd probably crush 7dtd's provider.

 

Non-smooth/blocky terrain - minecraft style - will make raycasting easier, because it is simpler to check collision against box than it is against system of metaparticles or a nurbs/bezier surface or whatever. Then again someone probably already dealt with that and there are optimizations available.

Write one.

 

makes sense. Can't wait to see an optimized version of planets cubed.

 

They renamed it to Stellar Overload. If you haven't played it since before your hardware upgrade you should give it another shot. Upgrading from a GTX 460 to a GTX 770 made it playable for me.

 

Nah TFP hired the guy, he works in house for them on the engine/terrain.

7dtd hardly make huge sales the game has its own issues, and its own gameplay, it doesn't appeal to all gamers. Unity providing a fully featured voxel terrain system that is flexible enough for game developers to utilize just opens up the type of games that get made. It doesn't crush 7dtd. I'm pretty much bored of static terrain games, and I'd rather play a good game built around a sandbox terrain system, over any game with static terrain, no matter how much more realistic it maybe visually on the surface, the moment you find its all just mainly static terrain, it falls flat. Personally see more interesting and enriching games being made with a good voxel like engine, than the effort being put into vr support.

I woudn't be bringing up things in this topic to discuss if I could just write one myself, stupid thing to say right.

 

Neat. Didn't know this game existed. Probably won't buy that unless they make it to steam store and (preferably) finish it.

In that case, Learn how to write it first or hire someone else to do it.

 

I never played it. I think it'd be a great stress test on the new hardware

It seems you both agree that's the best way to go about it, haha.

I'm biased so I can't say much about that other stuff. 7dtd is probably my favorite game right now. It'll probably be the best zombie game ever until someone takes the same idea and implements it better.

Zombie games in the AAA space... Look at all of the buildings and places you can't go in dying light. You do parkour over static terrain and when a demolisher shows up, the tiniest little plastic trashcan sticking out of the ground will cause it to bounce off as if it hit a concrete wall. The AAA polish on the static environment is still just a static environment.

 

AFAIK they're bringing it to Steam... Though TBH the fact they're using UE4 may have something to do with performance.

In any case, want to see great voxel engine? Play Blockscape, great game that is criminally underappreciated. Even better looking than SO and much more optimized (uses custom XNA-based engine).

I wish Jens would license his tech, because this is the best voxel engine I saw (better even than VoxelFarm as with VF you have to cheat regarding trees and place them as models, while in Blockscape they're voxels).

//edit: Regarding thread: I remember there was community-made smooth voxel solution, I've even posted in their thread about Blockscape to influence their terrain shader, but I keep losing the link.

 

They've stated in blog posts that they aren't focusing too heavily on optimizations right now. Plus Unreal itself is currently going through an optimization phase. Chances are performance will have improved quite substantially by the time the game is nearing its finished state.

 

Isn't the purpose of their project to try to have a really extreme voxel project in unreal?

 

I use the forums both as a Unity employee and as a regular user in my spare time (outside of work) where I do game development on my own.

This can sometimes cause confusion over whether I'm talking as a Unity employee or as a general user. But the clear wish from community managers etc. is that employees, moderators etc. don't use separate forum accounts for separate roles (and I do understand there's good reasons for this), so this potential for confusion is something we have to live with for now. Maybe at some point there will be tools to decrease the ambiguity.

Anyway, I was writing about voxel engines as a regular user with an interest in voxels.

As an employee have "lobbied" a few times internally for voxel solutions being built into Unity, argued that the first mainstream engine that gets built-in voxel support at the quality level of VoxelFarm/EverQuest Next will trigger a minor revolution in game authoring, but it's just not one of the top priorities at the moment.

Thanks for sharing your experiences with the various voxel solutions!

 

I had no idea there was such an interest in voxel engines still... Makes me ponder throwing mine out there for people to use.... its just sitting in a folder... collecting dust at the moment...

 

I think the interest is being roused once again because we're getting 'next gen' voxels. Minecraft is still unbelievable for what it is, but "stellar overload" is trying to do what minecraft dreamed of, haha. A voxel galaxy with multiple planets. It's going to be like minecraft + no man's sky... but with more variety in voxel geometry.

 

Sure. I don't think anyone would complain about having more options. Especially when it feels like most voxel engines become all but abandoned and/or only receive minimal updates and bug fixes. Or cost an arm, leg, and a kidney.

 

Maybe you could prefix your posts with some context, like "As a hobbyist" or "As a Unity Rep" because I read your post as though Unity were going to be adding Voxels some how into the engine and you were researching ideas and opinions for some kind of internal Unity project. Especially as it came out of the blue on a sleeping forum thread.
I've no problem either way, just it did get my hopes up that unity would do something builtin.

 

Are any there any voxel world oriented game engines?

 

I found one example of what's possible. A guy posted a video of a really cool minecraft-like infinite voxel world yesterday. He used UE4, blueprints only. It's the first time I've seen a copycat implementation done so well.

 

Wow. That and stellar overload are somewhat convincing me unreal isn't a bad option for it. But then again if I knew what I was doing, unity would be a perfect option too.

It seems unlikely but it'd be cool if there was an engine that was focused on having voxel primitives you could use to put anything together and build an open world.

 

By the way here is the Unreal 4 forum thread for that voxel game. It has some additional details like the fact he isn't generating a single mesh for each chunk but is instead using an instanced static mesh component for each block.

https://forums.unrealengine.com/sho...-an-Infinite-Voxel-World-similar-to-Minecraft