How many games should ship with dynamic GI for you to accept dynamic GI is a mainstream tech at the moment?
Also what makes you people think LPV isn’t dynamic GI? People always use worst case scenarios to justify lack of hardware power for running dynamic GI i.e VXGI in UE4 which is really horrible whereas Light Propagation Volumes have very tiny effect on performance and have been in shipped games since many years ago running on much older hardware. Why there’s no support from Epic Games for LPVs?
You are simply denying the fact that Epic Games in particular either doesn’t want to focus on dynamic GI or, they -can’t- handle it and neither of them have anything to do with hardware or the age we’re living in. It’s just how Epic Games runs their engine.
-Impossible to ship a game with SVOGI- … well there are games on the market shipped with SVOGI. Difference is the other engine developer spent some actual time on making it happen.
I don’t think there’s anything inherently wrong with them not focusing on it, especially with the other things they’ve been doing - such as mobile improvements, which I’d argue benefits a larger proportion of it’s users. We all know by now that Epic focuses on things based on it’s games first, and currently that game is Fortnite.
I suspect that DFGI started out as an experiment to see whether it would be viable or useful for Fortnite, but was canned when they realised they didn’t need it. Purely speculation but quite likely.
I understand the frustration for sure, but outside of the financial aspect there’s no reason someone can’t hire a rendering engineer to implement a solution that works for their game. For now at least that’s what it’s going to take if things like VXGI and LPV etc. won’t cut it.
That’s my point. Not focusing on a production ready dynamic GI isn’t because dynamic GI is some heavy tech for current hardware. It’s because they’ve been focusing on mobile and such.
LPV performs well performance wise = Epic doesn’t support it at all.
HFGI was fine = Epic abandoned it.
DFGI seemed fine as well = Epic abandoned it.
SVOGI was really cool but needed optimization = Epic shift+deleted it. —> Games can hit much higher frame rate than 60 with SVOGI enabled in Cryengine.
Debate of hardware, tech, market etc. not being ready for dynamic GI is invalid because it’s Epic Games that has been building a wall against any sort of available dynamic GI solution.
It’s good to keep in mind that most of the players don’t give a poo about visuals like dynamic GI and stuff. What most players care about is great story and gameplay. If you loved the Witcher series and then suddenly Witcher 3 came out built on Source (Half-life 2) engine - would you still play it? I bet you would )) I don’t mean that the tech has to be old, I myself would love to have my game properly lit and shadowed. But I’m more concerned about the lack of communication, features being abandoned a few releases later, not-so-great documentation and engine performance. This is what can prevent your game from being developed. Not the missing dynamic GI solution.
Regarding VXGI 2.0 and VXGI in UE4 in general - right now it’s not good for anything (games - poor performance, archviz/vfx - light leaking and so on). It’s a research project, not a game ready tech.
Why would you assume having dynamic GI would mean some other aspects of the game would lack?
You’re holding some issues in one hand and holding dynamic GI in the other and assuming if you try to get dynamic GI something will go wrong on the other side.
Can’t a company that makes 300 million dollars in 1 week afford to catch up with other engines in terms of dynamic lighting without ruining the engine on another aspect?
When people ask for dynamic GI or IBL or other common dynamic lighting features it doesn’t mean Epic should put pause their development on other tools and work on dynamic lighting instead. It means they should use a tiny fraction of their wealth to bring in some new people and try to make the engine catch up on dynamic lighting end as well. For example, physical lighting was released by 4.19 in a completely broken state and it was abandoned as soon as 4.19 hit the launcher. https://forums.unrealengine.com/development-discussion/rendering/1414326-4-19-physical-lights
Right, you shouldn’t lift your guys from what they’re doing and put them down on resolving some lighting issues. But my question is, why there shouldn’t be anyone at Epic Games who is hired and dedicated to working on dynamic lighting features and improvements at all?
After all, Witcher 3 with dynamic GI is better than Witcher 3 without dynamic GI and Unity now looks better than UE4 thanks to the continuous silence from Epic on all the lighting and rendering issues.
There is already a team at Epic dedicated to make this happen - it would be big,. That is their promise It is not just copy+paste, it has to be fast and it has to be production quality. . I cannot remember the thread… probably a few months back.
That’s the same thing we hear for years already. Well maybe this time ))
I’m sorry but you are getting it wrong. I was saying that there are things already that are more likely to cause problems for us devs than the missing dynamic GI solution. A car would drive without proper lighting but it won’t without proper physics. A multiplayer shooter would work just fine without GI but it won’t without proper networking. And so on, I hope you get the idea.
To be honest I don’t have a slightest clue what does Epics have on their mind. They make a huge move towards the film and vfx industry where high fidelity graphics are a must. But at the same in the Meet Mike demo project they use a hacky method of using a blueprint with an array of spot lights with overlapping hard shadows to imitate an area light soft shadows. Which you can perfectly see as soon as you swap the Mike model with something non-organic and smooth like a car or a mech. So why not make an effort and implement good lighting tools for a project they are promoting their engine with? )) I don’t know.
I remember when UE4 wasnt yet released to the public, an Epic representive told at a meeting for the third party studio, that “UE4 will have -only- DGI. there is no lightmass or such whatsoever, all the lights will be dynamical”. A year later he came again and told that “ok we have lightmass after all”.
Yup, there’s still a video on YouTube somewhere of Alan Willard demonstrating SVOGI using Elemental Demo assets (which IIRC, was running on a 3 SLI’d GTX 680’s at the time). They later realised it wasn’t viable tech for everyone (initially due to this generations underpowered consoles), so they dropped it.
It’s been explained a few times in the past. VXGI is a faster version of SVOGI anyway, so in the long run we haven’t really lost anything, other than it not being in launcher builds.
I seem to remember a lot of negative talk about the engine at the time, and how it had prematurely lost the “engine race” of this generation due to removing SVOGI entirely. Lol… look whose laughing now.
Nope. Actually they used to repeat “no active development on GI” like a mantra. They didn’t comment on LPV, etc.
Now it’s the first time they came to forums and proclaimed: “We’re assembling a team of lightning Avengers, working on GI, realtime raytracing and Lightmass”.
Showing off RTX demo - there was some collaboration with Industrial Light&Magic so we could speculate about huge deal behind a scenes to use realtime (or incredibly fast) GI for movies
Fortnite begun to produce dollars faster than federal money printer, so now it’s time to re-invest profits
I understand if someone would “annoyed” because we see progress on non-game tools, but not GI. Although… in near future we could gain a lot from Epic’s non-game initiatives. It makes a perfect sense to combine now:
RTX
needs of movie/archvis/VFX industry
needs of open world games or smaller games that could benefit from realtime GI
Microsoft hinted a new console briefly at E3, now it could be a good time to design a tech to support another line of “next-gen” games…
A movie industry used to develop techniques used later in games. Well, maybe now it’s finally time to work simultaneously on game and movie tech?
I’m pretty sure that rectangular lights were added to the engine because some experienced movie artist said: “Guys, why you don’t have this?”. I know, I know, we could live without area lights
I know what you’re saying but the point is GI has nothing to do with car physics and multiplayer networking and any other feature you name it. They’re different features handled by different teams so while XYZ might be 10 times more important than dynamic GI there’s still no justified reason for not having a rendering / lighting team whose job is to just fix rendering/lighting bugs and improve them.
Exposure isn’t physically based, Directional and Skylight aren’t physically based, there’s no IBL, DFAO controls are semi broken, SSAO is long outdated, Point and Spotlight physical values don’t work with manual camera exposure settings the entire dynamic lighting in UE4 has fallen apart.
Honestly it doesn’t matter what they said on the other forum threads. They’re silent on the threads they should’ve been responding to.
They are working on something, Juan Canada is involved. I’m pretty sure it’s dynamic GI, because he said it would be HUGE (I’m not sure anyone would consider improved outdated lightmap baking as something huge).
Dynamic GI really makes a huge difference in quality, I just can’t stand looking at these forests with pitch black shadows, or flat/grey surfaces that aren’t receiving direct light. It looks sad and dated. I understand they need time to make this tech, but some kind of teaser, or little announcement, would be amazing. Give us something to look forward to, I’m begging you Epic. :eek:
