[RELEASED] BG Light Engine (With Real Time GI Area Lights)

The GI in your early Sponza demo worked extremely well, and very fast and smooth on my mid-range machine... so it might be best (given the demand) to release the current version even if it isn't perfect yet. The only problems I could see in the Sponza demo were a few bugs with some of the shadows, but it isn't very noticeable.

 

It's your lack of communication! As I said before, I have a release Date and I relied on your promises. And you are advertising your GI. Directly on the first post of this Thread is a Video of your GI in Action, directly below is written in bold "Beta Version Available Very Soon!". Now you're writing here eventually. That is my problem!

That means I lost 80$ on you because apparently I won't get the Feature I bought it for or won't get it in time to use it properly. I'm a Indie Dev so I can't spend all my limited money on people who take it and then never finish their products like you do.

I would suggest you give me an estimated release date for the GI Release or a refund would be fair.

I'm pretty disappointed but I hope we will find a solution!

 

Well since everyone seems to want the GI in its current state, your wish is my command. I will work on an update and include the GI in its current state. You may have problems with it though.

Cheers,

Robert.

 

I don't really understand where all of this negativity is stemming from either to be honest. I've been following Brilliant Games and this forum for quite some time now. Whilst I agree his timeframe of the GI release could be a bit more accurate or informative, his communication has been pretty steady. I don't understand how anyone could honestly be asking for a refund.

He has stated numerous times that the GI is still in the works and is not available at this current time. What is shown on the first page is not so much an advertisement as much as it is a preview of what is to come. If you bought this product knowing full-well that it wasn't implemented yet, that is nobody's fault but your own. This is the case with any product, really. I don't see this being any different than buying an early-access or alpha-version game on Steam. The hopes are that the money you spent helps with development but you should be aware of the risks going into it.

Being a well-informed consumer goes a long way and so does a little patience. My .02

 

The negativity , mostly coming from my side, is reasoned in the lack of communication. I t was only when I asked that he told us that the project is on pause and the GI comes eventually. If I buy a Beta I expect continues work on the Product until it's finished, imagine you bought an early access game and the developer pauses the development. My understanding of that system is to give the dev a headstart not to get screwed and never see the final product. That statement before sounds a lot like "if nobody bothers, I won't finish it ever" and that's why I'm giving some pressure, because I want to make a game with it and have a given time frame. I can't wait until the dev finally decides to finish the product sometime next year. I think it wouldn't have been a problem if he had told us, that he works for the next two months on his crowdfunding campaign and the starts to finish this plug in after that. But he didn't and just said, I stopped working on it and maybe I will finish it, the money wasn't enough to keep me interested.

@robert: If the GI is not finished, I don't need it. My expectations where, that you finish this. I don't think that anybody wants this in an unfinished state. I bothered enough with this topic, I'm disappointed but it is OK, do how you think you should work but in my opinion, it is not professional and you shouldn't sell tools which are not finished, if you don't intend to finish them promptly. To be fair you should highlight on the first page of this thread and on the store page, that you paused the development for a unknown time and that the mac version is not working right now. Thats in my opinion the least what you should do, to be fair to future customers.

I will just buy the version from the Livenda when they are finished. At least they seem a lot more into customer support. Lesson learned!

 

Hey,

I was wondering how this lightning system performs on mobile devices. Especially on iOS. I mostly talking on replacing unity's directional lights and not so much about area lights. so it would be great if you could share any experience on that.

 

This will not work on mobile. I wouldn't waste your time/money if I were you.

 

Sorry for the delayed response, I have been away for a few days. As CaseJnr so politely put it, it is not currently optimized for Mobile. These are very advanced algorithms and as optimized as they are, aren't designed for mobile which is technically always around 10 years behind current generation PC's.

However, I will say that it likely won't run on Iphone, but it may run on others, how well it runs is definitely up for debate.

 

Just to be clear. It needs a bit more than just optimisation. It wont even run in OpenGL.

 

I would prefer if you could stop trying to sabotage this thread. This question was directed at the publisher. This negativity is uncalled for and the package does everything as advertised in its description. If you continue to spam this thread I will report you.

Thanks.

 

I do apologies if I have came across like I am trying to sabotage your product. That wasn't my intention. I was simply trying to inform Fred that product requires more than just optimisation in order to work on mobile. If I have said anything untrue or misleading please let me know.

Also I am not trying to be a bastard here but your message footer says that the product contains real time global illumination. It might be worth changing that.

 

Hi Robert,

I just wanted to say thank you for all the hard work and man hours you put into this project! Creating a GI system that is comparable to CryEngine is immensely complex please don't take the negative comments to heart! If they only knew how much personal time you had to sacrifice..more patience and compassion would suffice? I have been in an similar experience with a project! Wish all the best and looking forward to hear what you have achieved so far!

 

Thank you for the kind words. Yes it is very challenging. Negativity on the forums does not do anyone any good. It discourages me and it can give other people cold feet who might have other wise purchased it. Sales are the only thing that keeps a product alive. This product does exactly what's advertised. On the second update alone I put in 500 hours. I won't even begin to tell you how many hours went into the first update.

 

Hi Robert,

Not a problem thank you for taking the time to reply I would like to purchase the engine and support you, but I have 2 questions and it will be my main deciding factor before I purchase. Firstly can I use your engine in large open scene? I'm not referring to open world "GTA or Skyrim" for example, more focused but large in scale but not open world?

Think large open terrain outside mostly "Journey is good example of the scale of the levels they had if I remember correctly and to give you an idea of the look and feel I'm going for" Second question: I'm also looking into Time of Day - Realistic day/night cycle and atmospheric scattering asset for my night and day system will this work with your engine? I'm still busy playing around with the game mechanics in my head and testing in unity but these two parts play a huge role and are deal breakers for me. Thanks

 

Sorry for the delay on this response, I have been away a lot lately. The light engine has no problems with large scale scenes. However keep in mind there is no shader for Unitys terrain as of right now. Day and night cycle should work fine, though I can never make any guarantees when it comes to other shader based packages.

 

Thx Robert! I will need to check out the shader for the terrain. Goodluck with the project holding thumbs!

 

I can kind of see where the negativity is coming from but people have blown it way out of proportion,yes its bad news that brilliantgames wont be developing this for a month or two but theres no need to act the way ye are. Please continue developing this asset as I am only a user of unity free so I cant use other soloutions. Keep up the good work!

 

Apologies if I missed it in the thread. How would this integrate with Relief Terrain Pack? I find a lot of advanced shader packs have issues when I try an import them along side RTP, and of course, I don't want to get rid of RTP!

And another question : How much effect would this have on fully external environmental scenes?

 

Hello everyone. I just wanted to give an early announcement. I am working on version 2.0 of BG Light Engine!

It is a bit early to show anything off at the moment, however, I need some opinions. I have been working very hard on a completely Dynamic Global Illumination system and have made some ground breaking progress. Also, I have created my own custom shadow system that would work for all light types(including area lights) all while running on Unity free. While all this is turning out great so far, there is one problem. The whole system relies on dynamic for loops which unfortunately will only work on direct x 11.

My question for all of you is.. Is platform flexibility more important than quality? Would you rather have a fast and powerful GI solution with custom shadows, or potentially, something mediocre and possibly unstable?

Please let me know your thoughts everyone.

 

I dont use DX11, but for sure a DX11 solution for GI would be most welcome as option, maybe i decide to use DX11 later if it is guarnateed and tested for a PS4 release and all other systems work ok with it in DX11 mode.

What is the time frame for the release ? Also how much of the GPU power will this consume ? My game is already heavilly loaded and the room for any heavy special effect is getting really small, especially in combat.

Will it work in Unity free btw and will Pro have any advantage if so ?

 

Really cool. Have not been able to purchase this yet but am planning to. Especially now with the improvement you plan in Unity free.

My current pc does not perform great in Dx11 but I do prefer to work in it.

Send some info and screenshots as it progresses.

 

I think dx11 only assets provide the best quality (see jove 2.0) so I have no problem with that.

 

@nasos_333 I can't make any promises as of yet. But I will be sure to take advantage of Unity pro for the pro users. Obviously render textures will be an option which will speed things up.

@blueivy @Baldinoboy When it comes to real time GI, I just don't think its worth it anymore to attempt all platforms and older shader versions. Too many limitations that it ends up being a big mess. Because of how dynamic dx11 is, things can really speed up on a per pixel rate.

I want to share one experience I had today. I was messing with the for loop for gi sampling and accidentally made the loop run 1 million times. These black lines went across my screen and everything on my computer stopped responding. Not sure why I shared that but I thought it was entertaining that each pixel was attempting to sample 1 million gi samples on their own.

It's late and I'm rambling, I would love to hear more feedback on this DX11 only topic.

 

You can check the latest steam hardware survey to get an idea of how dx11only could be an issue.

Spoiler: For the lazy ones :)

From the top graph the results are:
dx11 gpu and windows 7 - 17.16%
dx10 gpu and windows 7 - 8.71%
dx10/11 gpu and windows xp - 2.86%
dx9 sm2b&3 - 1.32%
dx9 sm2 - 0.82%
dx8 & bellow - 0.15%

and the most used gpus:
Intel HD Graphics 4000 4.54%
Intel HD Graphics 3000 3.17%
NVIDIA GeForce GTX 660 2.57%
NVIDIA GeForce GTX 760 2.54%
Intel HD Graphics 2000 2.47%
NVIDIA GeForce GTX 650 2.05%
NVIDIA GeForce GTX 770 2.05%
NVIDIA GeForce GT 630 1.79%
Intel HD Graphics 2500 1.60%
NVIDIA GeForce GTX 550 Ti 1.59%
NVIDIA GeForce GTX 560 Ti 1.39%
NVIDIA GeForce GT 620M 1.36%

imho developing real time global illumination for directx9 systems would be an overkill, you could just keep compatibility for the dx9 systems though by having a minimum implementation of the system without any of the advanced effects.

 

I agree, if someone wanted gi while still targeting DX9 just stick to lightmaps if it's a quality implementation you want then DX11 means efficiency and quality.
Any eta on a screenshot perhaps?

 

I have 3 questions.
First, Does this support the occulus rift? You can download the free unity package from occulus website.

2nd, I am curious about the cubemaps/ reflections. Can they be box projected? Do they blend between each other if you have multiple rooms? Curious about how that all works.

3rd, Support for marmoset skyshop?


Also would love if area lights could have a texture applied to them like a neon sign.. i know livenda was working on something like that.

Oh and also dx 11 is fine

 

I think the data you posted is not the correct one, this is the one for the overall distribution of gpus (dx11 gpus are 74.03%):

http://store.steampowered.com/hwsurvey/videocard/

 

yeah i was referring to the top "DX10/11 Systems (Vista/Win7/8 + DX10/11 GPU)" graph. The one you linked is definitely better in this situation though, thanks

 

I'd prefer great quality with dx11,the majority of GPUs are dx11 now anyways.

 

Why would he support marmoset skyshop when this package already includes pbs shaders.

 

Because his might not have blending between skyboxes per location. Skyshop does.

 

Maybe ask him to try and implement that, certainly easier than getting an entirely seperate package to work with his

 

go with dx 11, go with quality choice

 

I'm a little confused. Are you referring to the current system and the use of cubemaps for image based lighting? Or are you implying that's how the new GI is going to work? But when it comes to reflections, I actually have another secret project... I'll give you a hint, ray-traced cubemap parallex reflections. I can't say this project would be included with BG light engine though. I was actually working on this before I started the GI. I have successfully blended and combined(combined as in, had each cube map fill in the other on whats missing) reflections with this system. But if it was released as another asset, I would be sure to make it easy to integrate with BGlights.

As a side note though, because these raytraced reflections have a great deal of approximations to speed things up, things like surface roughness are difficult to simulate.

Ah what the heck, I'll post a screen shot of some of the results I got with ray-traced cubemap reflections.(GI screenshots are coming soon). As for release date, it's far too dangerous to make any promises yet, as things come up in my busy schedule.

Edit: To answer your question on the oculus rift. None of this is screen based so there shouldn't ever been any problems.

Every curved surface in this screenshot is correctly reflecting its surroundings.

 

The reflections look excellent! Can't wait to see more on it
Is that the gi and area light shadows featured in that picture?

 

No these are ray-traced reflections using cubemaps that contain depthmaps. It's an experimental side project. Like I mentioned in the post, it may not make it into bg lights. But might end up being another asset that will be compatible.

 

Great work on the reflections, cant wait to see the GI

 

So w

Is it screen space reflections blended with cubemaps? That could be really handy.

 

Not at all. the process is similar to screen based reflections where it ray-traces to find the reflection. Basically its like a reflection probe where instead of using a box projection it actually correctly reflects what it sees.

 

Thats all well and good, but first the reason I intend to use marmoset skyshop is because it already blends between cubemap probes with full HDR, box projected, and roughness... Plus they have shaders like skin shaders, terrain shaders, other stuff.... Kind of why I asked for support for that since I am not sure you have any such shaders.

As for realtime reflections i plan on using livendas screen space reflections for that. At least for now on some of the dynamic objects.

I dont know how your current system works honestly. I found a tutorial thread that speaks nothing of how the reflections/ cubemaps work. So its hard to tell what your product actually does. Some videos would be nice. Also I am well aware of long development times that goes into making this sort of stuff... so i would not be looking forward to waiting many months for a product that might not materialize. Thats no insult to you, thats just the nature of development and peoples schedules in general.

Spectra GI was going to implement their GI with area lights as well.. there area lights are awesome and support textured area lights and stuff.. u should look that up and try to mimic that. Looked stunning. In fact heres a link to their youtube channel... I have no idea if they are ever going to even release since they promised end of july and then sort of just kept making more promises until the community closed their thread since it wasnt exactly coming soon...
http://www.youtube.com/channel/UCPpy0Nt_XfXnDl2LEx5piGw

Oh and the reason I dont know how it would work on occulus is one of your tutorial threads mentioned attaching a script to the main camera... well VR has 2 main cameras so how does that work?

 

Brilliant games had already implemented textured area lights far before livenda even announced spectra gi

 

@kurylo3d @JecoGames BG lights actually doesn't support textured area lights. Though it probably wouldn't be very hard to implement. As pretty as it may look, there's one thing he failed to show in the video. And that is the most important thing about Area lights which is often over looked. Angle attenuation. Reflections are the easiest thing to do in area lights(That's pretty much all I could see in his video), but diffuse lighting with proper angle attenuation is not.. I have never tried or seen any real time area light solution that does this properly, let me post these diagrams I made a while ago that demonstrate what I mean.


The bottom image is how every system I've tried has done it. It may not seem like a big deal to some people but this is what makes an area light... An area light. If you use this bottom solution, you will run into infinite problems. Surfaces below the horizontal line of the light will not receive any light, and not just that, there are many angles and areas where surfaces won't receive light when they should. People may not understand why this is happening, but your eyes aren't stupid, they know something is not right. The reason it's so hard to simulate this is because you are technically trying to calculate something which is infinite... After months of hundreds of algorithms, we finally came up with the most unlikely simple algorithm that simulated this perfectly.

About oculous rift, yes a script does need to be attached to the camera, but there are no render textures so it's no problem.

About screen based reflections. Yes there are some upsides to those, but I would never use them in my projects. I've actually made screen based reflections before and I can assure you, the limitations out weigh the benefits. I think raytraced reflection probes are the answer. Because you have the choice to make them static, and you can fill in the missing gaps with multiple probes. Obviously high resolution specular reflections with GI are a better choice, but it's very difficult to get those high resolution and stable enough to use for highly reflective objects. But it's something I really want for my GI solution.

Rant over. Hope everyone's having a good weekend.

 

Did you ever transition to using circular area lights instead of the point lights for your GI?

 

Yes.

 

Please make a texturemap able to be used on your area lights.. that would look incredable.. espeacially for neon lights and signs.

As far as I know occulus rift does actually use a rendered texture. Since it all is added to 1 screen... 1 camera per half the screen... And like I said its 2 cameras... one for each eye so which one gets the script? I would be happy to test it out for you, but not so sure I want to purchase just yet. My project centers around occulus rift.

 

Is there any noticeable difference between the two?

 

What he meant by render textures is that his system doesn't use any so it shouldn't affect the occulus

 

Yes. I have a pretty good idea of how it all works.. Trust me when I say, the only reason some packages have had trouble with the oculous is because they are screen based, meaning they are a post effect which is drawn on the screen afterwards which is usually built for one camera.. This is not at all how BG Light engine works. Its nothing more than a pixel shader which is automatically rendered on whatever camera is in the scene.

@blueivy Yes, it's mostly noticeable in lower resolution when the samples are more spread apart.

 

Real Time GI Using Light Propagations
​

Forget about that talk from me making this DX11 exclusive. It is now working perfectly in DX9. The way this system works is similar to Light Propagations method from Crytek.

Current Features:

-Full real time Specular reflections.
-Real time diffuse reflections.
-Ability to make any lights GI, static or dynamic.
-Real time control over resolutions and texture size.
-The amount of secondary light sources have little to no effect on performance.

This is all the info I can give away at this time. Enjoy this lovely screenshot showcasing real time specular and diffuse reflections. Keep in mind it is still very rough at this stage.

 

Is this still going to work in free? My understanding is that LPV requires a 3D texture,unity free doesn't have them