[Coming Soon] SpectraGI Advanced Real Time Global Illumination For Unity from Livenda

+1 to the GI first goodies later

 

+2 to the GI first goodies later.

I too check this thread way to often and have also been disappointed in previous impostor products. But Candela have proven they can deliver. This is a huge task they have undertaken so I think we all just need to relax and let them do their thing.

The letters GI do seem to create wars on this forum.

 

+3 for GI, terms, and goodies

 

i need only the GI

 

This is an interesting quote from Unity as they are integrating an 'existing API' which is primarily a re-lighting solution. Considering Enlighten they have their work cutout for them, it will not be easy, it is a fundamental change. It will have to work with thousands of Assets available on the store without the need for a DCC tool and don't break anything. 'Real Time' GI is not easy

We have more Spectra GI announcements coming very soon!

Livenda

 

Have you tested your GI solution with your wip water/ocean system? I am looking forward to both!! Did you use your SSRR with GI for your ocean preview?

"Coming Soon" sounds like a week or two..and seems to be never soon enough! "almost complete" is like you're almost there and then it will take some time to publish.. I don't know about anyone else but this phrase doesn't keep me on edge.. I can still be excited and breathe normal!

 

"...sounds like a week or two..." .. or a month or two.. "Coming soon" has been around for quite a while now :-/ . Sorry, Livenda-guys, I know you want to keep us excited about your product ;-)

 

I believe they said they're aiming for an August release earlier on in the thread. Not sure if this has changed since then.

 

Hummm...

[Coming Soon] SpectraGI Advanced Real Time Global Illumination For Unity from Livenda

This is a post I made earlier, people at Livenda... just say'n

 

I completely missed that somehow, thanks.

So basically this should be almost finished or in/near its final stages? If so I think I'd like to reissue my earlier question:

A couple months ago the reply was:

Hopefully there is more you can share now?

 

The silence obviously means they're cooking up an awesome video showcasing the amazing GI features we've all been waiting for

 

Question for you. I am waiting on my oculus rift dev kit 2 right now, but when i do get it. Will spectra and candella ssr work with the rift dev kit 2? If you don't know do you expect them too?

I just purchased candella ssr

Oh and do u ever intend on releasing your beautiful looking ocean?

 

As much as I want this asset to be released soon, i don`t think it will ever happen,,
A majority of the users being rather new to the games industry, some of them are,,, gullible.
If he had a dynamic GI solution, why on earth would he sell it on the Unity app store?
And why is the promo video just two static scenes with no dynamic objects?
Maybe, it could handle a small area with a few dynamic objects and characters. But,, that would be it, I think.

Seriously,,, Unity devs are working around the clock to get the enlighten working. As some of you here may know, Enlighten isn`t a perfect solution either. It is not as good as advertised. I`ve used UE with Englithen incorporated once.
While it does have a dynamic GI, It was PITA to get it looking good.
Their automatic UV unwrapping only works for simple primitive objects, so basically you have to make proxy meshes with UV maps maching the Original meshes`. But,, complex proxy meshes increase polycount a lot,,
It`s OK for buildings(You can make cubes and match the cube`s uv maps to the building`s uv maps, which ,by the way, depending on the complexity of the buildings, can be really time consuming and would require more manpower to get the job done).

Any complex meshes have to be lit from light probes. If you`ve used Enlighten,, you would know it`s very very difficult to do UV authroing of proxy meshes for complex meshes so their UV maps match the original complex meshes`.
And the baking takes,,, really long,, So, unless you have a render farm,, it`ll take forever to bake a large scene.
Even if you`ve baked a large scene with tons of objects, you have to divide it into small areas to lessen the performance hit. So,, basically,, you have to design your level so that when the lighting condition change in one area, the player won`t notice it in another area.

For a simple small game, enlighten will look great. No dobut about it.

Anyway,, if there was a perfect GI solution with 20 bounces that doesn`t need baking,,,
Why don`t they just buy the asset from Livenda, and put an end to their misery that they are going through.
(the devs are working under great pressure to get it working properly,, software development is just,,, pain,, a lot of times it costs more than expected,, and results are not satisfactory)

 

By the way, if you are looking to make a large game, I recommend UE4.
In their recent twitch video, they announced that they will soon switch over to Octree LPV renering pipeline.

Fable Legends will be the first game to adopt this technology( it is their in-house LPV solution).
Unlike Enlighten, it doesn`t need baking, meaning you can change your level design on the fly, and all game objects can be destoryed at runtime and have proper radiosity.

If you want a high-end game with blazing fast performance, go with Cryengine. Even UE4 is not in the same league,, yet.

 

@ HappyCoder84 Oh, it will come......soon.....very soon...really ;-)) *lol*

 

[QUOTE="
Anyway,, if there was a perfect GI solution with 20 bounces that doesn`t need baking,,,
Why don`t they just buy the asset from Livenda, and put an end to their misery that they are going through.
(the devs are working under great pressure to get it working properly,, software development is just,,, pain,, a lot of times it costs more than expected,, and results are not satisfactory)[/QUOTE]

You asked the question. But then that also asks why didnt unity buy a node based shader editor. Seems like a standard thing to have that unreal has. Why have they not bought any of the gui solutions. (actually i think the guy from ngui helped on their new gui but i just read that somewhere so i cant say for certain.)

Point is unity is making money off of anything side thing thats on the asset store.. they collect 30% already.. so why buy anything at all?

Plus asset store is an easy way to hook up with a lot of customers for the developer... might be a bit more dificult to sell it to someone who develops an engine without them trying to recreate or steal it first.

 

The anticipation for this thing is killing me.

 

That`s because,, something as big as this could be a real game changer. And if this was real, it would be pointless to keep doing business with Enlighten cuz this GI method is much better.
I bet they are bleeding a ton of cash just to get the Englighten working.
Overhauling the existing rendering pipeline is no joke. It`s such a massive project that no one knows how long it could take. They have showed us 2~3 promo videos about Enlighten, which leads me to believe that it is not quite ready for production.

 

There ARE benefits that Enlighten has over this. Benefit #1 is that, so far as I can tell, Enlighten will have better performance on mobile. The problem with fully dynamic GI is that it just isn't possible on low-powered devices - it's barely even possible on PCs, only high end gaming machines generally. That's why Enlighten is set up the way it is - so much of it is preprocessed, simplified geometry, runs multithreaded, etc. Because that was the only way they could get it running fast enough on devices with less computing power than a beast gaming machine.

There are also benefits this has over Enlighten, such as being much better at handling dynamic scenarios beyond just changing the position of lights.

 

I hear englighten has problems with extreme light changes as well.. for example a flash of lightning would look like bad, or a bright light out of no where.

 

Well, yeah. That's because it runs on another thread and converges over 3-5 frames, so really quick things like flashes will have some update lag and look bad.

 

This is a little off topic,, but does Unity have a dynamic occlusion culling like Cryengine or UE?
I can`t seem to find any reference to that. Can anyone shed some light on it?
Coming from Cryengine, I have a few concerns about Unity. I`ve tried pretty much every Unity desktop game. But none of them had a good performance, even though they didn`t have good graphics.
Some people chalk that up to developer`s fault. But,, none of those games seemed to have any complex AI system to cause lag. There is a game called Wasteland 2, made by some devs from a studio. They gained enough cash from their kick starter campaign to hire good programmers.

But,, the game`s performance was very dissapointing especially in some areas of the game. Same can be said for "The forest", "StarForge", "Might and Magic Legacy" and "The rust". I have 4 desktops, each with different graphics card to rule out any driver issue caused by Unity`s compatibility issues.

I`ve been using Cryengine so far. And I haven`t experienced any performance issues whatsoever.
I`m working on a FPS game(fantasy background, lots of vegitation, Complex AI system,,) using Cryengine.
Environment and AI part are almost finished,, So far, I haven`t had any performance bottlenecks. And all of the assets have high quality textures. Main character has 10 materials, face texture being 4K, the rest being 2K. Each enemy has about 300 animations, controlled and blended with 50 tags assigned at runtime from different systems, like perception system, locomotion system, or co-op Anim system.

That being said, I`d really like to give Unity`s new lighting system a try and see how it works.
My concern is, do you think the new Unity can handle such a graphically intensive FPS game?
When I play Wasteland2(which has been in production for a long time), all I can think of is,,Is this engine`s limitation or developers` fault?,, If even wasteland 2 devs can`t get it done, do I stand a chance?

Unity is a great engine. You can test fun ideas much quicker. And very easy to use.
I know no one has answer to this,, But I would appreciate some feedbacks on this.
Have a nice day~ Happy coding.

 

You'd probably be interested in this then

http://forum.unity3d.com/threads/jove-2-0-replacing-the-rendering-pipeline.230141/

He's reworking unity to look as good, if not better than cryengine while also doing performance optimizations. And their are some occlusion culling assets on the asset store. Hope that helps

 

The occlusion culling is not dynamic, that's a pretty difficult thing to do in a generic way from a technological standpoint. So Unity's occlusion culling is mostly static as far as I can tell.

Personally, I'm developing an FPS game and I really don't foresee any performance issues. There's some games Unity doesn't do well at - large open worlds like Rust are one of those things that, at the moment, are difficult in Unity. It's less for general performance issues, and more because some features of how Unity is designed make it difficult to do the right way. Which is a fancy way of saying "the Unity terrain engine sucks and so do the level streaming features".

Here's what I'm looking at for my game. It's not open world, and mostly static (lucky me), so as far as graphical features:

1.) Physically based shading. As this deals almost entirely with shaders, it has little to do with the performance of Unity itself, and more to do with how well optimized the shaders themselves are and the hardware they run on.
2.) HDR and tonemapping. Pretty efficient, not much to say here.
3.) Better bloom effect. There's an asset on the asset store called Natural Bloom, it's really good. It's efficient, and works much better than the builtin bloom for HDR scenarios. It supports a lens dirt mask, and scatters light in a more physically plausible manner.
4.) Motion blur. I managed to acquire Amplify Motion, which does the same motion blur technique as in Crysis (and BF, and UT3, and... well you get the idea). Again, mostly shaders so less to do with Unity and more to do with how well optimized the shaders are. I will say this plugin runs on mobile, so on PC it runs quite well.
5.) Ambient Occlusion is one I haven't solved yet. Unity's builtin one works.... but it kind of looks like a bag of smashed butt and doesn't run very well. I hear someone is working on an HBAO implementation, which may be what I need.
6.) GI is another one I haven't solved yet. At the moment, I'll probably go for lightmaps and light probe sets. There's this GI solution (maybe), and Unity is going to be integrating Enlighten. Either way, this shouldn't remain an issue for my game for long.

CPU wise of course there's very little my game is actually doing. Some movement code borrowed from Quake 3, some raycasts for weapons, and mostly lots of bookkeeping for scorekeeping and game logic and whatnot. Even the AI I have planned is going to be pretty smart but still really efficient (I'm going to be using HTN, which is perhaps even more efficient than behavior trees).

Really most performance issues in Unity come down to something you're doing wrong. Except open world. That's just hard to do in Unity period.

 

I think there is someone who made a dynamic occlusion culling method on the asset store.

 

I read somewhere that you planned on releasing this early auguest. Is that still the case?

 

Thanks for enlightening me, PhobicGunner
That was very helpful. Sadly, my game is open-world,, I guess I have no choice then,,
But, like you said, for your game, Enlighten lighting system will be a perfect match, I think.

 

Kurylo3d
Yeah,, I tried it just yesterday,,But, It doesn`t work great.
It only works well in very specific situations. I`m waiting for this asset to come out, too.
but,, I have a feeling that it will have some limitations(limited number of objects per scene, not fast enough for dynamic environments,,,)

 

Blueivy
Thanks for your tips. Jove 2.0 sure looks interesting. I`ll give it a try when it comes out.

 

Damn still no updates? Was thinking it would be released by now.

 

seems like nothing gonna happen with this GI... again

 

All these GI developers would go well politics with their broken promises.

 

Let's withhold judgement for a while, august isn't even over.
And its not like anyone has given them any money yet.

 

No but disappointing none the less — any news would be encouraging...

 

If I had a general-purpose GI solution that works for every situation,
1. found a company. Hire marketing people.
2. sell the license on a per-project basis to studios.
3.make hundereds of millions of dollars

Heard that Otoy`s octance renderer(formely known as Brigade) will be incorporated in UE4 in the coming months.
It`s worth checking out, though it is still far from being used in a real-time game.

 

it's been only three month since the trailer of this tool , i know we're all waiting for something, but this is not a little work.

i too expect some news/ videos of the GI system, but first : they said "well this is not set in stone however at this stage everything is looking good for a 'tentative' release date for late July or early August."
so we can count of late august, early september , maybe more .

Unlike other GI system like brillant games or DGI, they haven't asked us to buy an half finished project, in early state of developpement, we can expect here a finish product on release ( it will still need update, and will have some bugs , like all products ^^ ) .

I don't see why every one say that it will never be release, or that it is disappointing as it is not release yet and that this thread is just here to say us that they're working on this project.

this said : +1 to the GI first, goodies later ^^ , and even a vidéo with buggy things to see the state of the GI would be good ^^'

 

Awesome!
Looks like they are using a path-tracing solution.

 

To use the Brigade Engine 3 itself you have to have a High End GPU and even 2 of them as far as i know. But the visual quality it provides it's more than awesome. This is something i will love to see integrated in some commercial engine as soon as possible . The path tracer it's able to simulate almost every visual effect like MotionBlur, DOF, Caustics etc. It is able to do what the VRay and MentalRay renderers do.
http://brigade3.com/

If that's true i will think about moving to UnrealEngine course after i see wahat is the power of Unity 5 : )

And again if EpicGames intend to integrate this in UE4 will have to do some magic to get it usable!

 

Brigade is the "real time" version of the OTOY Renderer, and even on two titans it's a noisy mess. That's why I'm skeptical SpectraGI is actually doing any real path tracing here. If they were they wouldn't still be relying on their screen space reflections asset. But I'm sure we'll find out the real technique once they open up more.

 

They use some kind of PathTracing which is in Screen Space as one of the steps in the process. I guess it is all Post Processing not only the ScreenSpacePathTracing part and i'm pretty sure that such a method it's able to provide high frame rates. If SpectraGI is really able to provide high performance, many workarounds were needed for sure. Probably that's the main reason it is not yet released.

 

even a new estimate on a release date would be helpful... late july/ early august... has came and went.

 

They use some path tracing, but it's filtered and blurred (gi is low frequency). They can do it in a lower res, and upscale with an edge aware filter.
Brigade can't do filter because if blurring image, all details would be lost.

 

Hello Everyone,

Just to provide a quick update, we are progressing very well with SpectraGI, there are still a number of sections in the system which require thorough testing including further optimisations. Few things to note, the first version will ' probably ' not be viable to use on current mobile devices, however time permitting we would like to offer this option... so nothing definitive at this stage and No Guarantees.

Secondly, as we have mentioned previously, SpectraGI will have a number of methods available, which can be employed depending on your projects performance/ quality requirements including the trade-offs or gains (such as pre-computation time) during Authoring. There is a fast pre-computation step regardless which of the two main methods chosen. The Path Tracer is the 'third' additional option which can be used for ultra high frequency local bounce (when we mean local it is not as local as other SSDO implementation, Our SSPT covers a much longer distance) if needed and is not compulsory. From our tests, due to the quality of the other two main systems, path-tracing is usually not necessary at this stage. Please note the Path-Tracer in SpectraGI is our novel method and is 'screen-space' based unlike the OTOY mentioned above by blueivy. It will be quite some time before the OTOY solutions becomes viable for general consumer based hardware running at 120fps at greater then 1080p resolution with zero noise/pixelation, perhaps in a few years. Our SSPT is screen space and makes a number of assumptions to make it possible.

Lastly, the LPV solution being worked on for UE4 is not by any stretch a 'proper radiosity' system as mentioned by HappyCoder84 , LPV suffers from many issues (light bleeding is very distracting if not taken care of or at least minimized) however it is still a good method for small levels. The biggest issue is Performance! forget about it, and is mainly viable on DX11 hardware.

Regarding the time frame for the initial release; We are really pushing for end of this month, however No Guarantees, It is a complex piece of Engineering and we cannot just release a half-baked unfinished product (we would rather not release it at all if that were the case) given the response from the Unity community. We also have other products which will soon be available such as our project ' Total reflections ' , Nephilim Ocean engine which all have to work seamlessly with SpectraGI when released.

Cheers!

 

Thanks for the update Livenda, I'm sure this product will impress

 

Great news Livenda
Cheers!

 

why I feel so much better after Livenda updates? Thx

 

Thanks for the update Livenda. Keep posting substantive stuff as often as you can.

 

Thanks for the update — I'm sure we all hope it goes according to plan.

 

Thank you for the update! I, and many others, are looking forward to this. I can barely contain my excitement . I feel like its going to save a lot of time on lighting a scene as well in terms of man hours. Its going to be worth every penny.

 

thank you for the update

but where is our video

joke aside, great news for the estimated release date.