This is not really a valid comparison since the effects you listed have a far larger visual impact than dynamic GI. Going from flat colours to textures is hardly the same as going from no-GI to GI.
Then donât use normal maps. They increase ram usage. But Iâm pretty sure you do, because materials without normal maps donât look as good. New hardware arenât being built for 4K gaming only, but rather for pushing graphical details.
The matter of fact is, dynamic G.I is the new standard, almost every other engine already has a form of dynamic G.I. Give it just a little more time and the absence of dynamic G.I in UE4 is going to show as apparent as âthe material with and without normal mapâ.
I get the idea but I still disagree on the numbers. âA dozen AAA blockbustersâ hardly counts as ânew standardâ. The normal-map comment is also questionable at best because it is no big secret that the vast majority of gamers donât have PCs who can push dynamic GI; not something that can be said for normal maps. Maybe in a decade this will change but considering the emergence of VR has pushed up both fps and resolution requirements at the same time, I doubt it.
So youâre saying you donât see a justifiable difference here?
http://docs.cryengine.com/download/attachments/19377157/indoor2.gif?version=1
No, thatâs a gigantic difference, but itâs also the worst-case scenario. In a real-world scenario the effect wouldnât be nearly as dramatic plus youâd have level designers and artists do proper lighting meaning that the âbeforeâ pic would hardly be that bad. I am not saying that GI doesnât make a huge difference in âlaboratoryâ scenarios but if you take one finished scene, or a finished game even, and toggle GI on and off like in the pic above, the difference would hardly be that striking.
Iâm not sure what youâd call a real world scenario but hereâs an outdoor game world with/without very early version of SVOGI with FPS being shown on top left.
Thereâs literally less than 3 FPS difference here lol.
https://i.imgur.com/5lNOePT.jpg
https://i.imgur.com/Ou6SQ7z.jpg
Iâll happily concede that there is a difference at minimal performance costs. I just donât really care for it much. As I said a few posts ago though, Iâm not an objective judge here because I donât really care for graphics.
This is a flawed test because the non-gi version hasnât been run through lightmass. If you show the SVOGI vs the lightmapped version, the lightmapped version would look better.
Youâre showing a 3fps drop in a very isolated case with minimal geometry and no actors. Try showing it in a forest with a lot of trees/actors and get back to be on the fps drop. Iâd be willing to bet the drop is in the 10-15fps range.
You could probably make the lighting better if it were baked, but that type of game canât use baked lighting due to the size of the map.
But, for exteriors you donât need a complete dynamic GI system, itâs not as complicated to get good results for something like that. Switch to an interior with lots of lights and lots of indirect lighting and it becomes a different matter.
Whole argument about needing dynamic GI is due to not being able to bake lighting on 10x10 kilometers and you suggest to compare with lightmass? 
Thatâs Kingdom Come: Deliverance and that map is actually big with lots of objects. But thereâs some on/off test on trees if thatâs what you want to see.
https://i.imgur.com/I2QUpke.jpg
https://i.imgur.com/ypNakn6.jpg
Why am I posting these? just watch the video. https://youtube.com/watch?v=DctiRSFwoiIAnd again, thatâs old, SVOGI was still an experimental feature in cryengine running on older hardware.
I think part of the reason some of you guys oppose dynamic GI despite the fact that it makes games look better is (as you guys mentioned) that you think the effort and money is better spent on other useful features. That the money and man-hours spent on dynamic GI is a waste. But look on the other end, the money and man-hours are being spent on RTX instead.
Would you have been happier to have a working dynamic GI solution even if it cost you 15 FPS or have RTX that has nothing to do with game development? because thatâs what youâre going to get.
But if youâre coming from the perspective that you prefer them give you more smaller useful things than one big dynamic GI, well people have been crying for spline decals for years and -0- is the amount of attention they received from Epic. Thatâs a trivial thing to do for Epic. Man weâre still trying to figure out how to use/fix the âNew Physically Based Lightingâ that is shipped with 4.19 and theyâre not even responding to the posts regarding that being broken right after having released it to the public.
Iâm not sure where you guys are truly coming from but hey, eventually UE4 will lose itâs popularity when it comes to making games because making games isnât what Epic is developing UE4 for, theyâve been developing a GDC open world hype-only engine, a car showroom engine, a Paragon engine and now a Fortnite engine without taking others into account. Iâve seen the exact same situation with Crytek before their engine lost popularity. They made cryengine a Crysis engine instead of a game engine and everybody knows that. Now theyâre posting articles on 80.lv and trying to pull people back but they canât.
Basically instead of giving you a lego kit to make things with it, theyâre making a lego thing using a few pieces, and allowing you to disassemble and reassemble that as much as you want. When the entire additions you get to the engine are from whatâs been used in their internal games,youâre constantly limited to make games within the limits of their internal games. Thatâs not why people sign up here.
Time will make things right. But itâs going to cost a lot.
I personally only got involved with Unreal Engine because the many proven tools the Editor provides.
Realistic graphics were never my focus because once anything moves that ârealismâ is gone.
But GI if used right can give results, baking light isnât fun as already point out many times here.
Turning around again, a very good number of âFortnite playersâ simply turn off shadows entirely; they would do the same with GI, and not because itâs Fortnite, thatâs because they donât want to upgrade hardware. Ever.
ââââââThey wonât update their hardware for next decade or so until the machine is really dead⌠So whatâs the point really. The âgamersâ in the Gtx 10xx cycle are absolutely NOT majority and they wonât buy your game if they canât run it, to run it they will turn off GI, so again, whatâs the point?!
No, CryEngine never gained a popularity. There are literally few notable examples of non-Crytek games made with this engine.
It doesnât a minute to process the entire list
https://en.wikipedia.org/wiki/List_of_CryEngine_games
Letâs take a look at the Unreal Engine
https://en.wikipedia.org/wiki/List_oâŚl_Engine_games
And most of these game donât require a dynamic GI. Yeah, it could be useful, it would make development much easier, but most studios wonât look for another engine while starting the next project.
@Maximum-Dev
Iâm with you! I want any improvement to dynamic lightning possible, including dynamic GI. Still, I seriously doubt this engine would die because of this feature.
Especially that most of the studios behind open worlds are so large that they building their own engines. Ubisoft, EA, Guerilla or CDP donât care if UE4 has any solution for dynamic GI. Yep, I simply looked at the list of games you, guys, like to mention on this topic.
Someone mentioned Unity having the proper dynamic GI. Isnât it pre-baked GI too? And Unity still isnât able to handle asset-heavy projects. Even the simple asset management is a huge pain, something they gonna fix in 2018⌠Nobody would pull off Kingdom Come in Unit y.
Lots of studios abandon Unity after creating a first visually rich game.
That being said⌠Iâm curious how they gonna reinvest Fortnite profits. Maybe they plan to push for huge worlds in the next 2-3 years and eat the next piece of cake?
Epic never was a company forever focusing on the single game and forgetting about engineâs development.
Did they create shooter-only engine after the success of Gears of War? Nope, they created the general purpose engine, very convenient for designers and artists.
Did they support VR, mobiles or enterprise because of Fortnite? Nope!
Guys, you love to talk about RTX on GDC. Also, itâs very convenient to forget about Niagara presented on the very same GDC.
It did have a rise and fall though.
Look at the list again.
Cryengine 1 - 6 Games.
Cryengine 2 - 6 Games. (Crysis was a big hit. They invested a lot into making it a crysis engine, suitable for making more Crysis games as they did).
Cryengine 3 - 39 Games (Community was complaining about how the engine was developing at this time, including myself).
Cryengine EaaS - 17 Games (I switched to UE4 during that period).
Cryengine V - 10 Games (Some of it are indies with like 2% progress thatâd never release).
Had they listened to what people needed instead of only focusing on their internal needs today their forums wouldnât have been showing In total there are 12 users online :: 2 registered, 0 hidden and 10 guests.
Letâs take a look at the Unreal Engine
https://en.wikipedia.org/wiki/List_oâŚl_Engine_games
And most of these game donât require a dynamic GI. Yeah, it could be useful, it would make development much easier, but most studios wonât look for another engine while starting the next project.
I too, doubt most of those 2015 games wouldâve (couldâve) needed dynamic GI. But both tech and hardware has developed so much since then.
I didnât say UE4 is dying for not having dynamic GI. But the general response to community, whether itâs a rendering feature, a tool or simple QA has been lacking so much the past couple years it is eventually going to start hurting their business when number of games developed with UE4 decrease.
You gotta watch this: https://www.youtube.com/watch?v=vj1bUfSVBnY
On the topic,
Hereâs SVOGI OFF/ON running on a GTX 970 at 1920x1032. Very High settings. Difference is 12 frames here.
Thatâs a sample map not for production so not optimized and still if you disable volumetric fog/lighting, disable POM from ground layers, etc. etc. itâs pretty easy to get +70 frames with dynamic GI. My card is ~mainstream, getting low end pretty quickly. Steam HW survey shows GTX 1060 has been the most popular card for the past month and more people are upgrading as we go. If people have to turn of shadows in Fortnite to get desirable FPS and UE4 canât handle dynamic GI on top of that thatâs entirely something else.
Also note that in the Cryengine test above I have 7 shadow cascades and get to see the shadow from every pine needles on the ground.
People often lower the graphical settings in competitive games because it turns off visuals and gives them more of an advantage, itâs not because theyâre trying to keep their low-end hardware for as long as they can.
Thanks for your comparison. I wonder how SVOGI performs in interiors compared to a lightmap bake (quality and accuracy).
the effects I listed are just examples of scenes with good lighting (be it baked or dynamic) which look far more interesting than simply using a directional light and an ambient skylight (which is what you suggest as âjust fineâ)
again, if GI wouldnât be justifiable then Lightmass wouldnât exist
If GI wouldnât be justifiable, we wouldnât have hack in the form on Enviromental Mapping, Env Probes and Sky Lighting derived from HDR map.
Just sayin.
I definitely agree that baking is annoying as hell with larger outdoor areas. No debate there at all. My point is that SVOGI isnât some magic cureall. Itâs nice, but it has itâs pros/cons, just like any other lighting solution.
The game youâve been showing pics from seem to be using a very low end version of SVOGI, hence the slight drop in frame rate. The quality isnât amazing, by any means, but itâs still miles better than fiddling with skylights, PPVs and shadows; using the current lighting system in UE4.
Should UE4 have SVOGI as an option? AbsolutelyâŚ
Someone talked about Epic investing revenue earned on Fortnite to add new engine featuresâŚ
But arenât they hiring ~200+ people to work on Fortnite?!
Seems like all the earnings are flowing into making more Fortnite skins instead ^^
Nope, just checked and thereâs like some 30 open positions for engine programmers world-wide.