[Official] New Terrain System

Yes yes, exactly, it seems that .... they just wait for new assets, nothing needs to be done and the money goes ...
BUT, the asetss are different, and not always one person is able to maintain his product for years .... this should Be out of the box, and no matter how, I think so ....

 

I just found Unity update description on roadmap page about terrain system. But it is still in research column.
"Overhaul of Unity's terrain system with a focus on rendering of many hundreds of thousands of terrain objects, a tool set for fast world building, foliage placement, natural landscape decoration tooling."

 

well after all this year they finally have target for what kind of terrain system they want to create,
.
.
.
.
or just an update the description, but hey we got an update

 

Oh yes!
Hopefully this is a sign of "Work in Progress"

 

Would like to see: Voxel terrain, better detailed mesh (3d plants) behaviour (Noise map movement with more parameters), maybe auto texture?, the ability to set the height to a minus Y coordinate, better detail placement (Stones, maybe some leaves or sticks of wood. Not at all automatically placed, but as some kind of paint brush).

Water, though, should NOT be part of the terrain! I would still like to be able to place it seperately.

 

I'm all for unity, but things like these are what sometimes make me feel cheated.. 'hey look we're building an awesome new feature which you can soon try for yourself' and then nothing for years.. and the terrain system really does need an upgrade. You often can't rely on AssetStore items when they are complex systems and building something yourself doesn't seem worth it, when there's a constant background noise of ~soon it'll come oficially anyway~.
Please, this needs to be work in progress atleast by now.

 

Please Unity, give us an update about the Terrain System.

 

At least would be nice to know the plans, so one can consider if it makes sense to buy stuff or not.

 

It's just a hunch, but I have a feeling that some of the newer advanced terrain rendering systems we see in the Asset store might get bought up and included in future Unity versions. Because it seems like any new systems coming directly from Unity are still 2-3 years off. I could be wrong, but I don't see any other way they're going to get anything anytime soon any other way.

 

my hero for asking the question but alas

 

Wow that video shows they are not doing anything with it at all. Reminds me of when my Mom used to ask my brothers and I if we started dinner yet. We look at each other, chuckle, and say we are working on it but do not have detail.

'We realize it is a problem' but what I heard is 'Screw you we just want to talk about better cut scenes!'

That lady is my hero too! When she walked out I was cracking up. Wish more of the audience walked out on them at the same time so they see what a problem it really is.

By the way the Awesome Technologies page was updated and has more detail. This page made a tear drop from my eye:

https://www.awesometech.no/index.php/home/vegetation-studio/components/touch-bend-system/

This looks so amazing. Have not seen a large scale tree support sample yet but as long as it does the dev(s) of this is my other hero!

I love Unity and the Asset Store but just because of their lack in the care for a new terrain I wish this Dev would not use the Asset Store. Know that sounds mean but hate the idea of Unity to profit from a system that they HAVE TO support! How would people feel if Epic dropped PhysX and instead waited 10 years to wait for a USER to make a practical physics solution themselves and then profit from it even though they did nothing to help their own customers. Just do not understand it.

 

I am really surprised and deceived by this behavior.

In the past Unity was having several complaints for not replying to community real problems but I thought all this had changed...

This is so much not how I thought the new Unity - developers relationship would be.

Do not forget the OP was Unity, we are all here to support and understand... I believe a clarification from Unity on this subject is not only due but would also be proof of respect for many developers who put time and trust in the company for many years...

 

What I would demand for Unity's terrain system is not about content or high-level tools. I think that it is covered by far in the Asset store. I think instead that Unity should provide better shaders for ground and plants, geometry optimizations, procedural texturing, tesellation, etc. In summary, low-level tools that allow developers take their tools to a next level.
I guess that the goal of a terrain system is to achieve accurate geological features and biomes with a mouse stroke. If a good balance between the engine and the tools development is achieved, I'm sure that we will pretty near of that objective.

On the other hand, we don't really know how many of us are especially interested on terrain (photorealistic and accurate terrains in my case). Probably we're not critical mass for a big investment on the terrain system, but seems like products as Gaia, World Builder, Map Magic etc are quite popular according to the forum posts, so I'm sure that any improvement in this direction would be appreciated.

 

Yeah. On one hand you are right, but all this products have the same struggle and limitations as we all.
Because they all built up on top of the UnityTerrain System.

I wonder why there is no Terrain which is built from scratch.

 

well the answer for that is on the video above. .

 

100% agree! Building on an already broken system is not going to give great results. Believe Terrafirma and Awesome Technologies are both building their terrain system and vegetation system respectively outside of the garbage Unity Terrain system. At least I hope so.

 

i get the feeling that the terrain team are waiting for those tech to be mature then buying it :/

 

That would be great but I really do not think they will. They make a lot more money getting 30% of the sales from the Asset Store than buying it up. I would love these systems to be built in to Unity but believe they will be Asset Store extension packs. Either way we get great extensions! Just disappointed in Unity's lack of care for their program and their community. Just comment here and say anything. Even if it is 'We are sorry but at the moment there is no progress but we should be able to start working on it for Unity 2018'

 

With so many talented terrain developers out there I wonder why they did not think about making it an open source project as they did for UI.

I think would benefit everyone and at least we can see some progress...

 

They put the project on hiatus, technically it's still worked on, but the team has been re affected, they wait until some hardware are phase out from market to unleash a great fully featured engine not bog down by legacy process, but hey that's a theory, A UNITY USER THEORY, thanks for reading.

 

Who are you?
Indie dev

What kind of game are you trying to build or would like to build?
Top-down action RPG

How does terrain fit into that?
I use terrain for outdoor maps, and square tiles for indoor dungeons.

What use-cases, features, workflows do you have or would like to see?
-Have a material palette object that feed into the terrains. Currently you must build the terrain palette manually and if you change the look of one terrain it doesn't fix it everywhere it's used. For example if I want to tweak the smoothness of my dirt then I must manually repair all my terrains one at a time so they dirt looks the same everywhere.
-Same options that materials have. Such as changing bumpiness, allowing Standard (Specular) shader, detail maps. Currently you are forced to use Standard (Non specular) shader which is dull... https://forum.unity3d.com/threads/standard-shader-dull-compared-to-legacy-bumped-diffuse.413019/
-Automatic alignment of terrains at seam. Currently I use a script to change the offset so that terrains that meet up with eachother do not show a seam. Also should match height or other parameters.
-Allow resizing of terrain without scaling internal data. For example if I need the terrain to be a little bigger but I don't want it to move everything that I've currently drawn. It would either have to add additional data nodes, or interpolate between the old and new sizes to the new nodes.
-Removal of detail object limitations (If I remember right they cannot have scripts, an object hierarchy, or bump maps?)
-Detail object randomization (for example when I place trees/rocks manually I use a script to randomize their rotation, color, size, and select from a list of ~10 prefabs).

 

I think thats probably a small part of it (eg DX9 being deprecated) but I think the much bigger factor is the number of new foundations unity are laying. For example it now seems plausible that when they make the new terrain system, it will work with scriptable render pipelines rather than the current rendering system. Thats not an excuse or explanation as to why so many years past since new terrain was first touted, but it does mean that the situation we find ourselves in with unity 2017 makes me happy to cut unity some additional slack until these new foundations mature.

 

At the same time they talked about some other things in the pretty far future and weren't really willing to talk about terrain in any shape or form, which gives me some doubt. Maybe its a case of when its done radio silence that many big gamedevs adopt, though I'd personally prefer a trickle of information, even if its a drip feed.

 

Indeed! Nothing I said was supposed to convey a sense that I am confident about how far along in the process with terrain in particular they are. All I can say with confidence about the timing is that clearly they are not at a stage where they want to say anything at all, and I certainly would not take that as a sign that they are close to surprising us with something. Yeah, they dont even want to trickle, so there is no reason to get hopes up, and they obviously dont want us to get our hopes up right now.

All I was doing really was looking for some positive that is a consequence of the long time without any terrain love, and I'm not trying to defend whatever has happened or raise expectations that it will be solved soon. The positive I found is that when I look at recent Unity versions and what we know is planned for the next release or two, I see lots of different foundations appearing that I expect a new terrain system will make use of one day. eg everything from new editor user interface stuff (UIElements) to scriptable render pipeline to, especially, things being done to jobify as much as possible and increase performance, use of multiple cores & threads, etc. Some of this stuff will help the new terrain live up to expetations one day. But like I said, dont take this positive the wrong way - if we had gotten a new terrain system & tools a few years ago it would have been rather good for morale and would likely make it considerably easier to read me going on about 'how a bunch of new stuff thats starting to arrive and is understandably the present focus of their efforts and public communications will help the terrain system & tools be great one day'.

 

Or to look at this stuff I'm going on about with other new tech from a slightly different angle, I think its quite plausible that they tried one or two terrain things internally and hit various issues and bottlenecks. Maybe some of the new tech I mentioned is envisaged to overcome these hurdles, or maybe there was some other story to whats happened with terrain internally this decade and I'm wide of the mark. Or a bit of both.

 

@elbows said a lot of what I was thinking too. In my most optimistic moments I ever consider they might have an revolutionary idea that needs all this new technology and they don't want to leak details until they are in a position to implement it. In my most pessimistic moments I'm glad that moving core technology into asset store bundles is likely to mean that 3rd party terrain systems can integrate as well as the official one.

 

The terrain system is probably one of the best performing features constantly getting "patched" by Asset Store submissions - and yet failing getting close to what one would expect from a truly modern system due to its core limitations and the challenges a one (wo-)man team (which many Asset Store authors are) faces - who often impersonates sales/support person, UX designer, artist and programmer.
I suspect that the Asset Store solutions that spawned around the terrain system have formed an ecosystem that pays a lot of bills at UT - consequently one can expect them to be extra careful with their "cash cow". Why change or reinvent something that fills your bank account with a steady flow of money? Will a new terrain system offer the same cash flow? How many features should it have at maximum to still inspire 3rd party authors to integrate shortcomings indentified by the community to create a new stream of income?
I cannot backup these thoughts with numbers - but Unity means game development on microtransactions for a lot of small teams. Something missing or broken in the engine? Go, buy an add-on or a fix on the Asset Store if you cannot do it yourself! This is their business model.
If it was my business model I would need to think about these things. And decide accordingly...

 

The price in the asset store don't seem to suggest an important cashflow at scale big enough to support unity aside from the asset store team.

 

It's not so much the price, but rather the amount of sales. And we're pretty clueless about those, so everything related to this is more or less speculation.

 

Where did I leave that damn tinfoil hat?

 

Even optimistic Fermi estimation don't seem to suggest a scale relevant to have a cash cow relative to the size of unity. But that's my take.

 

Who are you?
Indie dev.

What kind of game are you trying to build or would like to build?
Me and my team specialise in third person, big world kind of games. Action adventure/platformer kind of stuff.

How does terrain fit into that?
Constantly, building out meshes for all the terrain is a pain, would prefer to only doing it on set pieces or special case situations.

What use-cases, features, workflows do you have or would like to see?

-For the love spagat, please give us nicer sculpting tools. The tools on offer are nightmarish to work with at times, the amount of times I've wanted to plane off an edge I cannot even count. Hand sculpting stuff is just made terribly hard, I think the terrain would be much easier to work with if it had tools more akin to what you would find in a sculpting program. It would also mean that you would get less of those 'Unity terrain' curvy as heck terrains everywhere.

-System to deal with stitching multiple terrains together. Would be great if it was easier to split stuff up onto multiple terrains.

-Tri planar mapping.

-Decals... its not a terrain specific thing, but damn it would be nice to have decals in Unity, to make lots of cool puddles and such ON Unity terrain.

-Is it possible to have nicer occlusion? So the terrain could actually be occluded instead of just downing its quality at distance?

-Change the terrains height without flattening the whole thing. Its kinda noobish, but also something that comes up when a project takes a turn you didnt think it would, and suddenly realise, you cant dig down anymore... would be nice to lift up the terrain without demolishing everything.

Thats all I can think of off the top of my head!

 

There's a feedback about it:

Isn't the number in braces the indicator of how often it's been purchased?

 

The proximity of that data to the rating (stars) suggests that it is the number of customers who actually reviewed the product. In your case 74 reviews led to that 5-star-rating.
I found some talks about sales to review conversion ratios on Amazon forums - they can vary a lot. Some convert 50 sales to 1 review. Most seem to need more sales to generate 1 review. Up to 1000:1.
Let's assume something good like 200:1 for a niche product ... at 74 reviews that product would have been bought by 14,800 customers.

 

Yeah like Macro_Sperling said that is the number of reviews including the fewer written reviews.

 

Terrain is actually the #1 reason I'm kind of wanting to jump ship to UE4. I probably won't for a variety of reasons. But really. Unity terrain needs a massive, massive overhaul.

 

How does UE4 do their terrain?

 

You can basically create your own shaders for Terrain using their node system to make it do what you want. We'll just say its capable of being ... much ... nicer.

 

So, I actually don't think it's in Unity's best interest to try to make "the ultimate terrain system". Not because of some asset store money train (The whole terrain section combined doesn't earn a significant amount of money).

Rather, I think it's much more important for them to focus on the backbone features that allow different terrain systems to be built on the framework they provide. There is no one "right" way to do a terrain system, any system you build will have tradeoffs that don't match for every project. Hell, I have two extremely complex systems for terrain texturing in the asset store, each of which has different strengths and weaknesses, and that's just the shader.

If they build a terrain system primarily about design time creation of terrain, then it won't be great for runtime creation, and if they do the opposite, the runtime creation aspects will limit the design time creation potential. Do you use a height field, or voxels, or vector displacement data to represent your terrain? Each has different advantages and disadvantages, which in turn trickle down to the rest of the system.

Every choice is a tradeoff which may or may not make sense for a particular game.

What they have done lately though is add things like GPU instancing- this is a feature that makes it much easier to have a terrain filled with lots of objects, which is a desirable feature for any engine, terrain or not.

That said, I don't think the asset store developers can fill the hole of replacing the current terrain system either. Most of us write stuff in our spare time, and even with several assets in the top rankings of the terrain section, I'm basically throwing away money by working on this stuff. Large assets require a ton of support, and there's really no way to get paid for that support, so when asset developers tire of the constant support they are expected to provide for free they slowly abandon their assets.

Additionally, while I'm excited to see potential assets replacing the terrain system entirely, without deeper access to the internals of the engine I'm not sure how well that's going to work out. There's really no fast way to create the physics data needed to make that work under the current APIs, so I suspect those systems will have to conform to the Unity Terrain structure and keep a Unity Terrain around just for physics.

So what's the right answer? I'm not exactly sure, but I do know that every time Unity introduces a new core technology that enables new things to be created, we see a surge of interesting assets created on top of it. When Unity added support for Vector4 UVS in 5.3, I wrote Vertex Painter Pro to take advantage of it (and give it away for free). The inclusion of Texture Arrays in 5.4 sparked MegaSplat, which in turn lead to CTS and MicroSplat. The introduction of GPU instancing triggered Awesome Vegetation. So I would prefer Unity to focus on these things, because only they can change these parts of the engine as long as it's closed source.

 

FWIW, here's the things I'd be looking into rather than building a new terrain system:

- Optimized storage, loading/streaming of high numbers of instances. Scenes and prefabs are way too bloated for stuff like this. We managed to pack our scenes down by several orders of magnitude in our current game (5mb -> 41kb) by writing our own formats. We added out own APIs to submit this directly to the render engine. (I'll be talking about this stuff in my session at Unite 2017). But really, we should have low level engine support for breaking up a world into chunks, loading them in and out without hiccups or GC, and drawing most of it without instantiating game objects.

- Extensible painter interface for painting on things. I've written a vertex painter and terrain painter which use plugin-able brushes to determine what happens when you paint on things. This takes a lot of time, and most of it is the same boiler plate stuff everyone is writing and dealing with.

- Way to submit height field data directly to the Physics engine. This is really key in getting rid of Unity Terrains.

- Efficient construction of triangle meshes from height field data. Let us specify the parameterization for LOD/patching, which materials are applied to which chunks, etc. The current terrain system makes a lot of assumptions here, many of which make it much harder for people to customize the terrain system (Drawing terrain multiple times, can only have one distance shader, shader lookups by name, etc).

- Better support for "private rendering". We need to be able to tell a camera exactly what to draw with what lighting- and not have it cull the entire scene or be affected by the rest of the scene lighting. This would make way for better imposter rendering, and make it much easier for me to make snazzy editor tools.

- A better garbage collector. I know, seems unrelated right? But the Unity Editor UI regularly allocates a ton of small garbage, and I spend significant time optimizing it, only to get complex editor UI's that glitch as you move around them. Any type of loading/steaming/instancing engine is going to fight with this constantly.

 

I also don't think they should necessarily give us a whole new system with voxels, builtin open world support or similar, that's something for the asset store, but atleast rewrite what's there to fix the shortcomings it has since its first inclusion (painting holes!) and its performance problems (trees not batching/drawcalls, bilboard alignment etc.), and/or give us more basic tools. Being able to submit heightfield data directly, as said, would be a great way to enable custom terrain solutions for the asset store. I also tried writing my own and having to build the collision mesh was what made me give up as i couldn't efficiently work around the hickups when assigning them to a collider.

 

Thank you for your thoughts on this Jason.

In the meantime, i am having a lot of difficulty making up my mind on how to proceed with these things.
I started out with the "basic" way of doing things.
Unity terrain (and trees and terrain objects), static batching, dynamic lighting, realtime shadows, LOD groups, Occlusion culling ...
Then came profiling and i noticed things weren't perfect.
Terrain to heavy, batching not reliable, shadows.... All in all too many drawcalls.
So i start looking for solutions.
Bake lighting where i could.
Stopped using Unity trees.
Legacy shader on terrain, gpu instancing.
But i am not a hardcore programmer.
I made a simple system that combines gpu instancing with the culling group api, which works ok, but isn't perfect and gives me problems with culling and lods.
I am now thinking about how i could use meshes instead of a terrain, but i haven't figured it out yet.
Each terrain mesh would be different from the next, so instancing is useless, and to make manual LODs of all parts would be an enormous job.

So i am sort of stuck coming up with my own solutions.
But we have the asset store.

There are some awesome assets in the store.
Your assets are certainly among those.
Combined with the support you provide, it gives me enough security to go with your products.
But it not always goes this way.
I have an asset for AI stuff, that just doesn't work good enough. Terrible GC spikes, bugs and an author who has disapeared from the forums. (99% great reviews though)
It took me many weeks to get into that asset, and now i'm starting all over again. (coding my own system)
And that feeling of "starting over again" is growing over time in many aspects of development.
My progress is seriously stalled because i have to figure it all out all over again.

Now comes Terrafirma and Awesome vegetation.
Perhaps these will solve some problems for me. But i am pretty sure they'll provide their own set of issues. As you said, single developers, not enough time and so many different use cases.
It will take a significant amount of time to familiarize myself with these assets with no guarantees...

Anyway, i guess this is the luxury of fast changing tech.
But i do wish Unity would provide better tools themselves instead of leaving it to the asset store developers.

 

On the long list of systems that attempt to improve the terrain, I've came up with one that I think it's generally missing. Based on the experience with my previous 2 free system I've developed a new foliage system based on the flow of UE 4's foliage system.

You can check the forum here or an advanced video setup here.

Some of it's features are:

1. Easy setup and painting:


2. Painting anything with using your shaders not only grass and trees:


3. Paint on any surface not just terrains:

 

that looks like pretty fine stuff!

so i just want to add my 50 cent and give you a preview of my upcoming advanced terrain grass solution:

 

Can we get an ETA for this?

 

i submitted it to the asset store more than a week ago.
so it should either be available really soon – or unity has some mystic reasons to not approve it. in latter case it would take some more days. unfortunately.

 

meanwhile you can have a look its lighting and wind features:

 

New assets are currently taking about a month to go through, I'm on day 25 with 3 new assets right now..

 

Nice! Does it support painting too or is it just a rendering solution? Grass only or any kind of foliage (like trees details)? I can't tell from the video.

Anyway, I see that people are really keen on pinning the engine's grass issue down, since it is one of the last features that's really missing from unity. (and yes, the built-in visual shader editor too).

 

Sad but true. Back in Feb 2017 it took more than 3 calendar weeks for my new asset to get approved (content updates otherwise were almost immediately approved). I though it was for Game Developers Conference, but now it just gives me the impression that Asset Store is particularly understaffed or too busy to deal with the new store layout.