[NEW PATCH] uNature - GPU, Interactable Grass. Interactable Trees, MultiThreading and more!

The most important for me are two things, performance which should be similar considering amount of grass I have on the terrain and second thing is grass interaction similar to guild wars 2 has (close range, dynamic movement).

Can You confirm that it will be possible to achieve that in Your first version ?

 

Do you mean touch bending ? Yep

 

Uhmm well can't see why not, uNature it self isnt streamed, but it supports streaming terrains

 

Can we specify wind directions for the grass? Can't do that with the built in system.

 

At the moment no, but many people requested it from me in the past week, going to add it now as a last minute addition

 

Nice asset. Might test it in the near future!

 

Just checking in, did you get a chance to test RTP and uNature? Super excited to see the demo.

 

Oh thank you for reminding me, yep!, works well as expected (Any terrain shader would work as I dont use any of the terrain features )

 

This actually indicates to me that there is a high upfront cost to using uNature similar to what you might see with Deferred over Forward rendering. Essentially that means that if a project is really close to the target fps, uNature won't benefit them because they won't gain performance, they'll lose it. So I really hope you are wrong in this evaluation.

 

Yep, there are many gains using uNature instead of Unity's grass either way, no spikes when loading chunks, touch bending, 100% generic code and more.
uNature is faster than unity in most cases, but there are some cases when uNature and Unity will be even, at the end of the day there's no magic that can give you 9 prototypes on a fully dense terrain with 0 ms .

I will look into this more and return with a more detailed answer.

 

This contradicts your previous statement that Unity will win by .2ms. This is the part that worries me. Unity should never win or there is a performance cost that could make this unusable.

 

I agree, there are places where uNature still isnt fully optimized, I will delay the demo for tomorrow and make sure I optimize it as much as possible .

 

K i've been researching this topic and it seems like there were indeed some "performance leaks" thank you very much for making me check it :d will fix today and release the demo.

Also - I've remade the touch bending to make it a lot more light weight and not require unity 5.4 and up.

The only disadvantage is that its not per vertex, but per radius.

The advantages are insanely fast calculations and literally 0 overheads and also support for permanent changes.

I can add per vertex support back and you guys will be able to choose it but it will require unity 5.4 and up and 10 touch bending objects max.

If theres a huge demand for it I will work on a new approach that would allow you to use per vertex with less issues, let me know what you all think

 

In my opinion after release we can see if it meets our requirements so let us lay our hands on it asap and don't let us wait anymore longer

 

Yep I agree, will probably release a per-vertex feature on V1.0.5 or smth like that

 

Teasing you guys too much :/
But here's the new extremely-light weight touch bending mechanic (requires 5.4)

Also, heavy optimizations are being done, sorry for the delay with the demo & trailer, doing my best to release it the best I can. I wont add new features in the last minute again, I promise

 

That looks really good compared to the early efforts. I'm hoping this is a low performance cost because it would be very useful.

 

The bending looks a bit strange.
When the ball rolls from one direction through the flowers it looks like they bend to the direction from where the ball comes instead towards the direction it is moving.

 

From the unoptimized checks i've done so far, it takes around 3 fps to simulate 20 objects all at once ( in your area only so in an 80x80 area-ish, or what ever you choose so not too bad )

Yea thanks for mentioning it, It was a mistake of mine, I put the strength too low, fixed

 

As long as we can specify which objects cause the bending then that's great because I really only want to have this feature for the player character and maybe a few enemies within close range.

 

Hehe no point having this feature if you cant choose which object has it

 

Better safe than sorry so that's why I asked.

 

Its all good! sorry if it sounded rude/ smth like that, ask as many questions as you want
That's why I am here

 

Touch bending looks great, what about ETA, this week ? Say yes

 

Currently on schedule (suprisingly ) so no worries

 

Seems amazing, most interested in the interactable trees system cuz my game should allow players to chop down trees procedually, that means a prefab tree is huge and performance comsuming. Maybe with this new system I can use terrain trees to achieve my purpose, look forward to it.

 

Absolutely. Currently from what i've heard of the testers its very fast and modular

 

RC2 has just been released for my testers, it means that the release is closer than ever!!

Left to do -
1) Going super hard on optimizations (1-2 days at most)
2) Documentation (1-2 days as well)
3) Tutorials (I plan to make them when the asset is in the submitted stage so no problems there)
4) Trailer (80% done, just need to get new footage from the new demo, so around 1 day)


Now, I've also decided to NOT integrate InfiniGrass, AFS and DirectX11 shader and chose Map Magic for a few reasons:

1) InfiniGrass 2 isnt out yet
2) I've started to work on it and its 100% useless, my asset has all of the stuff they support (more or less, missing stuff can be added easily) and even more, so its not worth it.
3) The assets are trying to achieve the same thing I do, so it would be pointless to integrate it.


IT DOESN'T mean that you can't integrate custom shaders, infact there's an extremely easy to use API for that, I am just saying that integrating the assets listed above won't be as worth as it sounds.


let me know what you guys think

 

Makes sense to me. I was wondering why you'd integrate a competing product that does the same thing. Made no sense to me at all.

 

Yea I agree, the key argument on that was translucency but that's also getting added (most likely on 1.0.5)
same for directional wind, so as you said, no point in integrating them

 

100% agree. There can be only one (grass solution)

 

The touch bending demo you put up there, will it support on IOS and Android platform?

 

Doese this mean uNature can be used with MapMagic?
If so, it would be fantastic, considering mapmagic and voxeland for the nex version of my game!

 

Yes to both of you

 

Okay guys I know I promised I wont add any new features, but there's one feature that I must implement or the release will be a disaster.

Basically some of my testers are using huge world chunks (8km x 8km) which is fine, but uNature only generates one manager for the whole scene so it gets really heavy.

I am currently adding automatically generated foliage grids that will allow you to paint any where you want and uNature will adapt to it (similar to UE4, no need to mess with instances/ anything like that, everything is automatic)


This will result in around 1-2 days delay so not too bad.
But wanted to let you know anyways as its going to be an amazing feature in my opinion

 

No worries m8.
Just for my curiosity, how is this optimization going to work? Am I going to set a grid size, like a 2x2km or 8x8km in your case? Will it be a performance difference according to the grid cell size and the amount of cells set in editor? I suppose there's a limited amount of uNature power that needs to be managed wisely according to each project requirements.

 

Nope, my hopes is to make it 100% automatic, just like UE4, you draw where ever you want and the system is generating everything automatically for you.

But you will be able to set a grid size if you want

 

Are you including a custom grass-painting tool? If so, I have a list of things I would like it to include

 

Yes!, a really fast one, that supports run time visualization and drawing
And all requests are welcome!, what do you have in mind ?

 

cool, I'm trying to do the open world with streaming in my game.
do you need any more tester? I really want to try this massive grass on huge land.

 

Not atm :/
Maybe when I finish the foliage streaming

 

1. Can we have support for painting on selected textures?
2. Can we have support for painting multiple grass textures at the same time?
3. And can we have support for having as high a density as we wish? Or is it still restricted to Unity's detail count?

 

1) Uhm what do you mean ? (copy colors from underlaying textures?)
2) Yes! already implemented (same for erase)
3) Yep, can't see why not

 

With regards to 1), if there are two textures on the terrain - a stone texture and a grass texture, can you restrict painting to just on the grass texture?

 

ohh that's an interesting feature, might not come to 1.0 but will come on the next one

 

Also, you might want to put a link to the Trello in your signature so people can easily find it

 

Done. also updated the trello and added more detailed version plans

 

Terrain composer 2 has this feature for anyone with that. Very useful.

 

Elroy I'm hungry for more screenies

 

Haha no worries, the moment I finish the implemention of the last feature that I noted above I will post the trailer