It’s pretty easy to see why Epic hasn’t implemented a new, better, dynamic GI solution. Just looking at other games, game engines, tech demos, and tech docs, there’s 0 high quality, low performance impact, general use dynamic GI solutions without major drawbacks.
Enlighten isn’t truely dynamic because it requires baking, and the baking takes forever. Lots of leaking, seams, errors, artifacts. Also imagine it would be very expensive to license.
VXGI. Only works well for smaller scenes. Very demanding.
SVOGI, going off the CryEngine/Lumberyard implementations, it works fine for outdoor scenes, but falls apart on even a simple interior room.
My point is that i don’t request any new functionality I just want current implementation LPV bugs to be fixed. LPV worked as expected in 4.16.3 but not anymore. And it’s sad.
I’d like to say thank you to Epic that we have LPV as it is. It is far far from perfect ideal dynamic GI and it’s outdated technology - but it works with acceptable performance(comparing to VXGI - quality is perfect but performance…ehhh) and it is possible to tune it to acceptable render quality almost ever. So performance-quality relation is quite good. Of course we don’t have cascades in LPV and i guess many other things that could bring LPV on level up. But we have what we have. I understand that it is very challenging task to create excellent dynamic GI.
If Epic will finally bring new dynamic GI to UE4 - it will be fine for all of us
Nope. It is actually true and (I thought) well known fact in industry. Idk if that was ever official statement or something that Epic employees usually tell you. If Epic does not need or use feature internally they don’t develop it. Unless you are very big 3rd party licensee.
It’s actually bit disappointing, since their internal projects rarely align with what the engine is mostly used for. It now happen thta Fortnite BR pushed engine in direction that it is used by lots of people but IDK if that was accident or planned because well “people make open world multiplayer games, and we need to test engine”.
Upside of the situation is that features which are used by Epic are actually well developed, tested and proved to work in real game development, not proved to work with theoretical engineering (like Unity).
Yeh - a quick google shows I’m probably wrong about that. I could have sworn I was using them in UDK but meh, that was many years ago.
The tech and/or idea might have been created by Crytek, but Lionhead/Epic obviously implemented their own version. Doesn’t really matter…
I doubt Epic are going to do much with LPV, especially if they’re not using it internally and with it not being a fully supported feature. Like almost all GI solutions it only really works for limited outdoor scenes anyway. It looked good on Fable Legends but if what has been said in this thread is true (I’ve heard different but meh), then it was only ever designed for that game anyway.
I expect they’ll take pull requests though. Unfortunately it looks like we’re limited to fixing it ourselves for now.
let’s pretent for a bit that pull requests are actually a fluid process…
even if someone could fix it, submit a PR that Epic accepts and get fixes/improvements integrated, nothing guarantees it won’t break again because as Epic doesn’t use it internally, leading to the same situation that the OP had when creating this thread
It’s often true, but not always true.
The biggest Epic games aren’t designed for Nintendo Switch, mobile, VR/AR. Still they’re doing great job on it. Robo Recall was more like a way to “battle test” Unreal’s VR than a standalone product made for profit. Epic games don’t need big system like Sequencer too much, but lot of non-Epic games use it.
Of course all the tools for character animation are so good because all big Epic games use it heavily
There’s no awesome open world support because Epic doesn’t produce open world games.
There’s no game engine that will satisfy needs of every kind of game.
But saying that Epic doesn’t care about community and needs of very different games… come on… Did you notice difference between UE3 and UE4?
UE3 was great engine for every kind of game, but only if this game was a FPS shooter with few cinematics, not too much of “living world” elements and… if you was OK with that general look of the game would be similar to Gears of War. Gamers where quite to surprise that The Vanishing of Ethan Carter didn’t “look Unreal”.
It’s a bit different in UE4…
It is important that the sentence ‘if the Epic doesn’t use internally, then they dont implement it’ is true, because some people think Epic somehow is lazy (or any negative) and they start teasing around it. Because once the code is in (PR is accepted), then Epic will be fully responsible to support it across multiple devices be it Xbox, Pc etc. And there is no way to be fully sure about PR quality until you are actually using it in actual games played by real people (not just some games sitting in the lab). It is that serious.
And a hundreds of small studios, man. Especially VR, AAA studios wait until VR market would be ready for big productions.
Also… I’m pretty sure those “big partners” would love to have fully dynamic GI implemented for them. For their open worlds. And that’s not happening.
Is Niagara, new material layering workflow or new audio engine are systems only needed by “big partners”? No.
Most of the systems/features serve studios of any size.
It’s truly amusing. UE4 is one of two most universal engines for indies and I’m so often reading “oh, Epic cares only about big partners”
Ok, readed the theme, and i think some of you people are very wrong
LPV is a good already builtin technology and it’s worth to develop it, because the performance of hardware are increasing rapidly and with it’s expanding you dont need ideal GI - you need sufficient GI for your hardware/ And LPV is alredy here - give it some tweaks\features(like cascades) and it’s enough
Its like UVLayot softwere - 10 years old and still one the best in 3d industry.
Give people what they are asking - LPV, cascades. and with increasing of hardwere productivity - it will be enough for time being. And then you can think of new GI for as long as you like.
LPVs have always been an experimental thing and probably won’t be iterated on any further. There are quite a few other real-time GI solutions out there and they all have their ups and downs. Pick your poison, preferably one with more frequent updates, and work around it.
Well the good news is Fable 4 is under development by MS, and for all appearances they’re back using UE4 (LPV was originally being developed for a different Fable game). Maybe they’ll try to commit something to the main codebase again. Then again maybe not. Other studios that use UE4 are baking their own GI solutions without putting it in the main codebase. For example I wouldn’t be surprised if Obsidian’s new open world RPG (uses UE4) wouldn’t end up with any commits back to the main engine, despite the fact that they’ll obviously need something better for such a game.
Oh my gosh… I’ve been searching for hours to find a way to make the emissive material work with LPV in 4.19. Couldn’t figure out why it’s not working anymore… I’m so sad. Even Unity has it working now! I might as well go to Unity, ■■■■ it! My project is at halt!
Any project that crawls to a halt because it’s missing cosmetic features is arguably a project that needs some re-thinking either way. Sorry if that sounds rude but I’m really tired of people claiming that features like these are of GROUND-BREAKING IMPORTANCE when a) 99.99% of those same people don’t have the technical know-how or resources to actually make a game with AAA realistic graphics that would justify the need for, in this case, dynamic GI and b) hundreds of AAA multi-million dollar successes released and sold without dynamic GI. I have never heard anyone say “Oh gee the gameplay of this game is amazing but I can’t really recommend it because it just doesn’t have dynamic GI”.
By all means, please go and shift your project to Unity, their entire business model is advocating fancy features that have never been proven to work in the real world and most people don’t actually need but lack the engineering experience to realize it.
I’m just angry because now we are left without ANY solution for unbaked GI. And I’m angry because Epic IS still using it in Fortnite and Paragon, so tell me why they ditched it from UE4? It worked fine until 4.16, now it doesn’t, and doesn’t seem like they will bring it back… So I cannot make animated TV light, animated lamps, or any animated sci-fi glowing materials, no help to light interiors without HOURS or even DAYS of baking lightmaps. Not to mention that making lightmap UVs for UE is a nightmare, since it’s one lighmap per mesh and not per material or UV set, so i have to split meshes to tiny pieces. Also, seams are visible on lightmaps, so it’s yet another obstacle to keep in mind…
Unity did a massive improvements in regards of lighting. They are atlasing the lightmap UV islands into separated lightmap textures, so you can make bigger meshes without splitting it to small pieces. They added the progressive rendering, which is fantastic! The light probes are very cool too, since you can tweak the GI density/precision manually in the scene if you spot some GI bugs. And then finally they implemented the emissive dynamic area lighting I mentioned before.
It has screen space and planar reflections, has HBAO, volumetric lights and shadows are now much better than ever. So it doesn’t miss anything that UE4 doesn’t have. Actually there are things that I wish to have in UE4, like proper root motion and humanoid system, fbx objects hierarchy with proper pivot points, non skinned mesh animation, or reusable between scenes curve/key animation system for any kind of object.
The reason I’m using UE4 are blueprints, integrated AI, best material editor I’ve ever seen and Terrain tolls. There are obviously many many more reasons, like fantastic community, great support and so on, but as an artist, for me bringing the assets to UE4, animating and lighting is a nightmare. Epic should seriously think more about it.
I just want current implementation LPV bugs to be fixed. LPV worked as expected in 4.16.3 but not anymore. And it’s sad.
