I disagree. With GI everything looks better, nearly instantly. Lighting is softer and more natural. That’s why there is static GI in the form of Lightmass.
Some types of games simply cannot afford to use static GI.
SVOGI is hardly beyond cutting edge, as the latest iterations of CryEngine shows it possible to optimize it to run even on consoles and it adds a lot even in open levels. Especially with a lot of foliage, which looks far better with GI.
It’s really not about the fact that the feature is to much strain on hardware, because it is not. It’s only matter of fact that Epic does not need it for Fortnite.
Because almost all (if not ALL) open world games are now using dynamic GI. It looks much better, and I think it’s fair that people expect to have good graphical features with today’s technology. It’s not even a beyond the cutting edge feature, since most other games engines have real time GI. I’m pretty sure I’m not the only person here who got tired of spending tons of time baking lightmaps, kepping his fingers crossed there wouldn’t be a small leak somewhere, forcing you to bake the whole scene again. It’s outdated, it takes too long, and it can’t be used in larger scale games with tons of foliage.
Graphics can be tweaked in the option menu and most features can be turned off to run on any potato.
Better graphics does not mean worse performance. Cryengine runs much faster than UE4 while having tons more graphical features than UE4 has.
Implementing graphical features isn’t just about putting in some code, call it a day and stop supporting it i.e vol fog/lighting. It’s more about continuous try to improve it, making it run better while also looking better.
It is quite realistic to expect every single high end feature to be present in this engine. At current rate, UE4 is in transformation to become “the potato game maker machine” because industry has been moving forward and Epic has been moving side ways.
But Epic has proved to not be in the business of providing production ready/optimized features i.e recent so called physically based lighting addition. So any discussion from GI to physical lights to open world tools to something as simple as spline deals for making roads is quite pointless.
summed it up pretty well.
Is it needed for Fortnite
Is Fortnite needed it
For Fortnite is it needed
Replace Fortnite with whatever their next game is called and you’ll have a clear understanding of how much improvements you’re gonna have to expect.
I think Epic really only have one guy who would be working on something like this, that’s probably one reason.
It would be a huge help to have GI since it would make development much faster and allow you to do more types of games. The games that are fully dynamic are usually using some type of low-end solution that only works in the situations that they’re using it in. For UE4 to add it, it has to work in more situations, you don’t want to design a game thinking you can use fully dynamic lighting and then find out later that you’d have to bake lighting for one of your platforms because it can’t handle dynamic GI.
Not one. They actually hired quite a few rendering engineers and researchers. Like 4-5. And opening Rendering r&d department.
The point is they don’t need it for Fortnite, and the Rendering resources were mainly allocated to real-time Ray tracing which is cool shiny feature. Totally useless for next 10 years, and everybody already forgot about it. But it was nice show for GDC.
Look the other thing worse that we are not getting svogi (when it is proved to work reliably), is the fact that we are seeing, that existing features do not get (seemingly) attention like shadows optimization, or when they get attention, they break everything with no explanation, like unified physical lighting.
Look not complaining about poeple who are working on Rendering. They have their tasks, and they are doing them. I’m only complaint about those,who create those tasks.
More or less, all unrest revolves around dynamic open world features.
Terrain rendering budget in stock UE4 is comparable to a frame budget of a game. On that background, dynamic GI talks are pointless pretty much.
Those are fair points but even for open world games I’d argue it’s a fringe feature. The two most successful open world games of last year - Breath of the Wild and Nier: Automata didn’t have a hint of dynamic GI and it didn’t really hurt them at all. My point is that dynamic GI won’t make or break your game and people are treating it as the ONE feature missing for them to achieve their dream of a gigantic open world game, when it’s not. It’s extremely far down the list of things that need to work even.
You can’t really say dynamic GI isn’t of much importance. People are using static GI because they don’t have access to a fast/good looking dynamic GI. Who would like to wait half day for baking a scene if they could just use dynamic GI? (that could also be tweaked for viz quality renders like in cryengine). And lightmass is really not a solution for outdoor/open world games as it takes 65 million years to bake/iterate/bake. Dynamic GI makes it all look better, let’s you iterate and see the results instantly, tremendously cuts down production time and doesn’t have the limits of static GI. It’s only drawback is, Fortnite and Paragon didn’t need it and that’s why Epic is letting LPVs die as they did with Heightfield GI and DFGI. GDC showfloor was all they cared about, sadly.
Quick question: which of CryEngine-based games using SVOGI run smoothly with full details on? 4K or Full HD only? (Games that did use renderer of the vanilla engine only). I’d simply like to see some examples from the market. I ask because I ignored a lot of AAA recently.
Of course, as a designer, I’d love to have truly dynamic GI in every engine! I don’t need to discuss it. Even in UE4 which isn’t designed with open world games in mind.
But is already possible to do deliver it today without killing game’s performance? I don’t see a clear answer for this question in this topic (perhaps I missed something).
Miscreated
Kingdom Come
The Climb (I guess)
I’m not sure about others as I haven’t been looking there at all. But if it wasn’t for crytek being broke, I’m pretty sure there’d have been more games.
“The performance depends on which GI settings are used. Usually on Xbox One it takes 3-4 ms of GPU time and on a average PC (GTX 780) it takes 2-3 ms (AO + Sun bounce, no point lights, low-spec mode). The fastest configuration is “AO only” mode; this provides large scale AO at a cost of less than 2 ms on Xbox One.”
^ That is ~1 year old.
GTX 780 is pretty low end GPU atm.
Additionally just download the engine, load the sample forest level and enable SVOTI and see how it runs first hand.
I can say that it’s not of much importance because no game ever will have its make-or-break point on dynamic GI. Breath of the Wild has for all intents and purposes just a skylight and a directional light, which anyone can use for open-world games and it sold millions. Would it have sold more if it had SVOGI instead of its barebone setup (Assuming hypothetically that Switch could pull it off in terms of performance)? I honestly doubt it. As far as any real-life game production is concerned, dynamic GI is a nice thing to have that can be used as a boasting asset but to say that any game suffers greatly for the lack of it is I believe ignorant at best.
You’re using Zelda as a reference point to prove dynamic GI wouldn’t have done any good. That’s fundamentally wrong as Zelda is a heavily stylized game and doesn’t rely on realism. But there are games that rely on realism. Take GI from them and they look as ugly as Donald… duck.
Nier Automata had a realistic design and didn’t have any realtime GI either. The only open world games that I can think of that had dynamic GI are the new Far Cry, Horizon Zero Dawn and The Witcher 3, and I’m not even 100% sure on the last two. My overall point is that dynamic GI won’t make a bad game good nor will lack of dynamic GI make a good game bad. I argue that in the grand scheme of things, not a single quantifiable thing about a game - reviews, sales, concurrent players etc. will be affected in the slightest by dynamic GI. Actually scratch that - I believe it can only affect it negatively due to increased hardware requirements. This makes it, in my judgement, a useless feature when it comes to game development. If anything it’s juts another “GDC showfloor” feature to cater to the archivis and offline-rendering crowd.
The lack of GI won’t make a game bad of course, but you still need it to make an open world game up to today’s graphic standards. Horizon and TW3 are using GI btw, just like all open world AAA games. Nier didn’t have GI and it showed, with dark and flat lighting and shadows. Good game, but looks very dated.
And how would it affect a game negatively? It works with all engines. Cryengine has it, had it for years. And it runs better than UE4 without any GI. Unity has it. All AAA in house engines have it. The “increased” hardware requirements is not an excuse.
I suppose that’s the case, and I admit that I’m not the most objective judge here because my hair stands on edge every time I hear the words “graphic standards” or “graphic improvements” or any other variation of the term. If you hadn’t noticed I couldn’t tell if TW3 had dynamic GI nor could I have noticed that Nier: Automata didn’t (I inferred it from it running at 60fps on the PS4) and I don’t see how Nier: Automata looks dated at all. In my mind the pursuit for the nextest and bestest graphics is what ultimately drives game budgets into insane figures and dooms games to repetitive, safe “AAA copy-paste trash” that big publishers are pushing out on a yearly basis, hence why I find every “revolutionary” graphic improvement such as dynamic GI as ultimately useless.
I’m not a pro or something like that but as an amateur all the open world games i’ve seen in unreal engine 4,they were looks bad in terms of lighting,the last one is state of decay 2 i think.
I’m not really getting tired of watching the lighting and jungles in a game like Kingdom come,it’s just lovely ! and well don’t know if it have GI or not.
Of course epic always shows beautiful tech demos with UE4 but at one side no one can be as good as them in using their game engine and at other side Tech demo is Tech demo and game is game…
heh I wonder why Lightmass was invented in the first place if flat lighting is totally fine. or why need PBR if Phong shading is totally fine. why do we need normalmaps if only baked diffuse is fine? why do we need textures at all if you can get away with flat colors?
The workflow and iteration improvements dynamic GI would bring are at least if not more important than the visual splendor. Having to wait minutes and often hours for every change to lighting just isn’t a productive way of working. If Epic wants to stick with Lightmass, I’d at least expect them to incorporate things like a progressive GPU lightmapper similar to what Unity and Frostbite have or better compression algorithms to that we can blend between different lighting scenarios to make dynamic time of day at least possible. My point is, it doesn’t even need to be fully dynamic GI but Lightmass at the moment simply lacks a lot of features that would greatly improve it’s usability.
