A Unity ID allows you to buy and/or subscribe to Unity products and services, shop in the Asset Store and participate
in the Unity community.
Discussion in 'General Graphics' started by bibbinator, Jul 4, 2014.
Which would you suggest that not only look great, but outperform current Unity terrain solutions?
You are new to Unity ?
lol. Nope. I wanted to see which magical terrain tool performs better than Unity terrain. I haven't found any yet and I own quite a few.
For example, RTP looks great, but is so slow it's unusable for me.
Or move to another engine more appropriated if you don't find any plugins good enought for you.
With a bit of optimisation and decrease some distance values RTP should work good, i don't know what you are trying to do ?
RTP slow? How did you manage to do that? : )
Easy, use it on a console system like Wii U. Totally unusable even stripped of most of its features.
Hi magique i think this project is interesting
Eager to test it
Check out also this
So ppl are still making game for that console? Thought it was abandoned ages ago...
Now that's what I'm talking about. So there is a magic tool out there to smash Unity terrain performance. But does it work with SM 4.0 and without DX11 Tessellation or other DX-only specific features?
It is a mesh based terrain so i think so
Sounds good. I tried the test program and it seemed to do well on my PC here, but performance differences on the Wii U are not always so dramatic. I've posted over there though and will continue discussions of it with the author. Thanks for the tip.
MegaSplat will outperform Unity terrain texturing, depending on the features you enable, and runs on anything which supports texture arrays (shader model 3.5+). It can splat map 256 textures with just 3 samples per pixel per texture, and offers packing modes to allow you to pack more into less textures and save additional samples. I even have NPR lighting modes coming in the next patch which allows you to use texture based lighting instead of PBR calculations, or disable normal maps for extremely low-end rendering.
I haven't tested on a Wii U yet (no dev kit), but it should work..
I've been eyeing MegaSplat, but not quite sure about giving it a try yet. I have some reservations.
because i expect a decent built in terrain system, that's it
because i expect a new vegetation system to build my assets and store assets, with my custom shaders too.
not everyone has RTP and who will maintain it in the future?
and since they spoke about a new terrain system, we expect some news... what else
Yeah totally agree with @antislash. Yes there are decent systems developing for terrains, specifically for vegetation, but a built in system would still be much appreciated. Especially since they were advertising it here some time ago.
As an Asset Store environment dev I have to plan on users using the scene in default Unity and that is tricky because default Unity in its current state is not going to work great. Yes I will make compatibility for a dozen other Asset Store tools that will make it work better but it is ridiculous to say to the user 'I understand it is not working for you so that is why I have a list of other Assets that you HAVE TO purchase for it to be usable for your project'.
So basically I feel most Asset Store Extensions should be a great addition, boost to your project but not a necessity that you can not finish a project without. Especially in an engine as popular as Unity.
Thankfully since Unity is not stepping in there are a few systems that are saving users now like Critas Tree System by @Assembler-Maze and AltTrees by @Diab1O . It would be nice though to have a working base in Unity that these systems can only make better. Otherwise it is like Unity is just a leaky roof and the Asset Store is Home Depot. Should be an addition, not a repair.
You should also mention LushLOD Trees by @AaronBrownLM and uNature by @ElroyUnity
I can conclude you don't have a game needing advanced terrain, otherwise you would have already pick up another choice.
And being builtin don't guarantee that it will be maintain in the future, they do deprecate stuff.
i also use Cryengine.
i'd like to know what will be the unity terrain system specially for assets, but i wish i also can use it for gamedev, depends.
i also use unity for archviz, idndn't buy RTP just for that, but the actual terrain sucks badly.
if i didn't need an advanced terrain system, i wouldnt be in this thread.
Oh sorry, i misunderstood.
For those interested in workarounds (till we'll have the official fully working unity solution), my workaround SpeedTree rendering 'optimizer' can be found on the asset store here.
@Andrew_Bowell Any news on this subject? It's been quite a while...
Thanks in advance
I'm surprised there isn't a terrain beta, as lots of other systems are being reworked via public beta, Input/Physics/Navigation/Rendering/Timeline.
Would this be a good way to introduce a new terrain system?
Because there is already what you need in the Asset Store, as always Asset Store solutions are better than Unity specially for desktop games.
I'd tend to disagree with you. I still can't find a terrain solution that can give me a vertex distance of 0.3 meters and a splatmap resolution of 512x512 per each 400Meters of terrain (without splitting my terrain into multiple difficult to manage chuncks). Or maybe I'm wrong?
just assuming,it still not in a good state to be released for public use (even it is beta). ie if you check the Adam exterior scene they have a custom terrain system (with texture array, tesselation etc) but still custom component not a native stuff, and sure confusing as hell to use
at least a small report on what's been worked on to keep us informed
Except when you need to update your project to a new Unity version which breaks an asset (5.6 breaks multiple terrain assets, for example) and are at the complete whim of that assets developer to fix it. That, and there will always be stuff (like the instancing support that Unity recently added, and I'm sure many more upcoming features of the inbuilt terrain system) that could never be achieved through an asset.
Asset store solutions are only better until Unity decides to tackle a particular thing, be it UI, Post or hopefully within the next 10 years - terrain
You mean when Unity contract the best asset store solution provider to update their systems e.g. UI?
Didn't the guy who did that Visual Node Based Shader solution start working for Unity, what happened there?
Well UI, very little of it if any uses NGUI. It did initially but they wanted something else. And the final result does not in any way resemble NGUI.
I believe he is busy on Unity's post at the moment? Let the man speak for himself @Tim-C
i'm really curious on how this project going tho, especially since now unity have more people who've been working on AAA game title and engines. :/
Getting silly now. We're almost another 3 years behind (since this thread started), and nothing to show for it other than 12 pages of hopes and wishes.
Ffs Unity, get a team together, and light a rocket under them.
Yeah progress is actually kind of slow compared to a typical AAA game which also supports multiple platforms. I guess that the switch to subscription will at least allow things to be released when done.
i guess Terrain System is a myth now
Yeah, it's sad that my Breath of the Wild Killer will have to be put on hold (alongside my WoW and COD killer) because an existing terrain solution doesn't fit my needs (like sculpting and texturing)
You're right. Why won't people see that brush based terrain editing and lerped terrain texture blending are just fine for their projects? Why would they like Unity to come up with a new system instead of cobbling together a system from various assets from the store? I'm completely baffled.
On a more serious note: I completely agree with this. Additionally, combining different assets from the store to cover the various needs that are currently underdeveloped about unity's terrain tend to make terrain setups quite fragile.
i am waiting to know how i must produce vegetation assets for own projects and store.
at least what should we be prapared to ?
will the terrain support custom vetagation assets or will it be 100% speedtree (that would suck)
how will we produce grass assets ?
will the terrain textures need a particular setup, what will be the resolution supported ?
will we have access to splat maps to produce custom shaders ?
will we have a road tool,, a water tool ? and so on and on....
unreal is tempting to me atm...even if i will loose unity c# capabilities
I went from 25 - 30 FPS on Unity terrains of a real world road 21km long with very little detail. I used Real World Terrain to grab the geospatial coords for the terrain heights and satellite maps. I used EasyRoads3D v3 to create the road system from OSM data and flatten the terrain where the road is. I used the Terrain To Mesh plugin and broke the 6x6 2km sq terrains into 1/2km grid (16 each terrain) and used MegaScatter to throw detail vegetation along the side of the road out to 150 meters and combined the meshes. This is with wind and full shadows. I am now getting over 600FPS with 100,000s of vegetation detail models, including forests of low poly trees. Once I finish the culling algorithm my framerate will shoot back over 1000. Umbra culling dropped the framerate by several hundred FPS. I went through two dozen or more plugins to get this combo that gave me these phenomenal framerates and spent several months testing this and that and the other. Not optimal. But now I have the pipeline i can do what took me months in about a week. Unity is probably over-thinking this. A simple switch to accomplish what I did semi-manually after terrain setup would do the trick for at least 50% of the cases.
oh wow, i thought it's only me who prefer to go with terrain meshes rather than using built in terrain.
Does Tesselation work on 'land' meshes?
Technically it should, depend on the shader.
Yeah, this is getting silly. Unity is the most commonly used engine in the world right now and the company is most likely making several billion a year.
Just take some money and straight up buy out a terrain solution like they did with TMP.
It is simply a mesh. It should do fine on appropriate hardware. I just added thousands of SpeedTree trees and am still honking at 250+FPS. I did combine all the separate LOD Materials for each LOD level replacing each LOD with the materials from LOD0, which dropped bathing calls down. I am still trying to figure the difference between Static Batching and Instancing for performance and what i lose and what i gain with each method. Next up is figuring out the new lighting system ad how to bake it into all the detail object without accumulating 100's of megabyte of lightmaps.
Thank you @ippdev for this real-world use case. Astounding results. I've been wondering if a move to sliced meshes would improve things but I didn't imagine the magnitude of improvement you've found. I wonder if Unity devs can take note and maybe get in touch with you to chat through what you're doing.
$90 Real World Terrain
$45 EasyRoads 3D Pro
$25 Terrain to Mesh
$50 Mega Scatter
$210 Total for a working 'racing' game FPS large terrain area with roads in Unity.
And I'm guessing you have spent more on trying other tools and assets?
thing is : the actual terrain system doesn't match with the possibilities offered by mechanim and C# . i mean, what's the gain, if i have to develop my own terrain system ? or buy third part plugins.
with no decent terrain system unity is like a racing car without tyres..
just got this thread ... nice one ...
Who are you? What kind of game are you trying to build or would like to build?
I'm Antonio, game passioned for 40 years ( I was 6 ) with solid technical IT background. I joined unity last september after 20 years spent in IT Financial domain, and I hope to continue my new journey for long time. Planning a VR Game in a procedural endless world.
How does terrain fit into that?
My world is going to be endless, covering different landscapes, mountaints, deserts, sea, etc, and all the seasons ...where is it ? Somewhere in the space and time
What use-cases, features, workflows do you have or would like to see?
I would see caves merging automatically standard meshes having same Interface for both
End of November I have started to develop an asset to generate automatically routes on a terrain connecting multiple points and find the best route according to terrain morphology... it works and beta is going to start soon .. plan is to release that by end of April / Middle of May on asset store ...
I spent about 500 bucks over a year to get this set of tools. The issue with me is that each Asset Store terrain system and splat map maker or shader is different and requires getting up to speed with. I would get one for shaders and then the vegetation would not be double sided or my framerate would drop significantly. I would get another and it would require painting vertex colors. Get another and still have crap framerates I nearly gave up several times and canned the project..but I am doing it for business partners..so..