The Finals used RTXGI but I’m wondering if there are differences here. Arc Raiders doesn’t have destructible environments, but it does probably have time of day and maybe weather.
Unreal Engine 5, [UE5 Developer Availability 2022-04-05]
- Thread starter mpg1
- Start date
The Finals used RTXGI but I’m wondering if there are differences here. Arc Raiders doesn’t have destructible environments, but it does probably have time of day and maybe weather.
I assume that's just a general bucket for continued Lumen performance and scalability improvements. Is there a reason to believe that has anything to do with Arc raiders?
Yes I imagine it will do something similar. There doesn't seem to be a whole lot of GI in the gameplay trailers but RTXGI (can we rename this like... probe-based indirect diffuse or something please) is often pretty coarse in the first place so hard to say for sure.
Frenetic Pony
Veteran
The comment about maybe going free to play reminded me of Lumen, and at times Nanites current performance costs. It seemed from the presentation like the goal at Epic is currently geared towards moving these features towards maximum scalability, there was mention of using the same art assets for nanite all the way down to mobile, and that new material system (I forget the name) has the ability to compile multiple material variations for different hardware targets while trying to maintain a similar look.
This seems like a big goal to achieve, trying to stretch the same lighting and art setups all the way from mobile to say the PS6. I'm just wondering if that's really an overall goal things are being organized around, or if it's more fragmented. It's not even clear to me if "Lumen on mobile" is a specific hardware target, or if it's just something aspirational someone put in a slide while general improvements are being made.
Lumen on mobile is only available when a UE project is configured to use the desktop deferred renderer for those platforms (in experimental status)
cheapchips
Veteran
Embark had a 'disproportionately high number of rendering engineers' three years ago. I think they're all ex-Dice? They could go in whatever direction they like for rendering. They've made their own performant RTGI engines for fun it seems.
That they've stuck with UE as a core for their first two titles speaks volumes I think.
They got lost in Rust. Happens to the best.
Frenetic Pony
Veteran
Yeah, but the slide I posted from the video also had a goal being "Mobile Lumen" for "future versions". And the Nanite portion mentions trying to get Nanite to run on mobile as well. As fortnite ships on mobile, and there's currently 2 versions of Fortnite assets and lighting setups with mobile and console/higher, it'd make sense to try and unify that down to 1 target.
With recent research trying to use shadow maps and/or screenspace traces for many shadow casting lights even megalights could, potentially, target mobile. It seems an interesting but ambitious goal. I suppose for Lumen one could go probe GI only, totally separating screen and temporal complexity from GI. You'd be able to guarantee that it runs both incredibly quickly (you can cache probe hit points and only relight most probes most frames) and in constant time, at the cost of light leak, laggy GI updates, low detail diffuse, and no detail at all in specular. But that could ship on mobile (and go well with Fornite's mostly diffuse art style anyway).
The Rust project is another thing entirely and always was, it's a long term "make your own game" style project that's so popular with kids today. They've been slowly working away on tools for things like applying physics to player generated characters, and then applying procedural animations to those physically simulated player generated characters. I.E. stuff UE isn't set up at all to do right now (if it ever is)
While that's going sticking with UE for "networked shooter", I.E. fortnite like, makes a lot sense.
Last edited:
Not a fan of Lex Fridman or his interview style, but there are some UE parts in here. I haven't listened to much yet. I kind of skipped to the UE6 part. There's not much, but what he talks about is Epic is slowly building up their Verse programming language in Unreal Engine for Fortnite, and it will eventually find it's way into the standard UE for game developers. Seems that's a big part of UE6 having the Verse programming language as a big part of the engine.
There's a bunch about UE and it's technology earlier in the interview, and I'm just listening to that now.
Does he explain why they dropped Mega from Epic MegaGames? I always hated that. They should bring back the Mega or at least add a Turbo or something.
Me too, but the people he gets. Well, I guess I somewhat disagree about not liking his interview style. There is plenty I don't like, but I like the long form format that lets people talk about subjects as much as they want.
The demo was impressive in 2013, but many ps4 games have surpassed it, like spiderman.I remember Epic demoing the UE4 engine at the February 2013 reveal of PS4 and no subsequent games looked like that.
But the fact that this PS5/UE5 demo it is doing so much more than that PS4/EU4 demo: insanely-detailed models, amazing lighting, vast draw distances, inherently better baked-in animation and high speed traversal through a huge open space with lots of detail and things happening.., Yeah, that is showing a lot of real, deliverable potential. This isn't hey look how shiny, it's hey look how shiny and and it's all working in what is similar in scope to sections of Tomb Raider and Uncharted.
Again, I think this bodes well.
Last edited:
some interesting qtidbits that he published independently.
How Unreal Engine was first created | Tim Sweeney and Lex Fridman (26 minutes)
From 5:42 on he explains so many interesting things. How UE started (I had the first Unreal 1 game on PC, amazing graphics at the time), and how they almost went bankrupt several times and how they stretched Epic's finances nearly to the breaking point.The first version of UE required 20 people, a HUGE team at the time. They barely survived. They were on the verge of running out of money many times.
Tim Sweeney on John Carmack | Lex Fridman Podcast Clips (2 minutes, quite interesting)
Last edited:
The whole thing was pretty nice. They went through the early years, his start programming, formation of epic, unreal, some specific vintage tech sweeney himself developed, modern unreal, future unreal, code parallelism, fortnite metaverse, apple legal troubles, steam, epic games store clunk, exclusivity contracts, Verse programming.
One of the most relevant takeaways for this forum was probably when Sweeney comments on how most of the rendering code of Unreal has been completely re-written multiple times, while a lot of the gameplay/simulation code is a lot more stagnant. He recognized the single-threadedness of UE's game logic is a major weakness and a major focus for UE6. He later talks more about this when discussing scaling and verse.
One of the most relevant takeaways for this forum was probably when Sweeney comments on how most of the rendering code of Unreal has been completely re-written multiple times, while a lot of the gameplay/simulation code is a lot more stagnant. He recognized the single-threadedness of UE's game logic is a major weakness and a major focus for UE6. He later talks more about this when discussing scaling and verse.
@milk Im very interested in Verse. I’ve never done any functional programming. The language looks and sounds like a big departure from typical procedural or object-oriented programming.
I also really like how they’re developing it alongside UE for Fortnite so they can have it broadly battle tested before moving it into the main engine.
I also really like how they’re developing it alongside UE for Fortnite so they can have it broadly battle tested before moving it into the main engine.
Unreal Engine is insane when used by the right devs. Let's be honest.. the amount of crazy things you can do with the engine is staggering. It gets a bad rep for having a "visual style" but we all know the reality is that you can make all types of visuals in UE and everything inbetween... One problem is a lot of developers seemingly use it in the same or similar way.. which leads to sometimes samey results.. compounded by the fact that people can literally just go and use freely provided assets, you get what you get. But in the hands of the right devs, it's incredible.
Games are only as demanding as developers make them, so I don't blame games using advanced rendering features which require lots of power. The main issue for me is the hitching which comes in most games these days. I think Epic have done a good job so far addressing the shader comp issue, and more devs are definitely doing it.. however the traversal stuff needs to be ironed out now. Games should guide you smoothly through their worlds.. so hopefully the engine improves to fix level/asset streaming and hopefully more effort is put in by devs to iron out stutters.
IMO we have to get this stuff under control before moving on to anything else.
I’m more forgiving of shader comp stutter as the engine is trying to work around an underlying issue with DirectX. Traversal stutter makes no sense to me with todays massive system ram pools, SSDs and fast PCIe.
