Game development presentations - a useful reference

Lumen is hardware accelerated now too, so other hardware accelerated GI solutions can still compete.

 

I mean, embark is making its own engine and providing a shitload of stuff free and open source, so yes there is a point. Not to mention, Lumen will not remain as the one and only GI solution forever.

 

Of course there’s a point. Lumen is far from perfect.

 

Not to mention you can only use it with UE. Unless all game engines die and everyone uses UE (which there is kind of a trend for) there will always be space for other solutions.

Also, this appears to be the way forward for the embark renderer. Surfels + ReSTIR + luminosity cache (like Lumen). UE isn't the only company with good engineers.

https://gist.github.com/h3r2tic/ba39300c2b2ca4d9ca5f6ff22350a037

https://github.com/EmbarkStudios/kajiya/tree/restir-meets-surfel

 

If you're about 100,000 research/development-hours behind Lumen you're going to do what, exactly?

 

If you couldn't make as good a gi solution as lumen, there would still be many reasons for a studio to make their own engine and their own renderer. There's lots of good reasons not to use UE.

 

If a game is targeting 60 fps on PS5 and XSX for example you are not going to want to use Lumen even if it has 100,000 dev hours behind it.

 

You're making a case here for no-one to ever compete with entrenched giants. For Red Bull not to get into F1 because McLaren, Ferrari, Williams. Smaller can be leaner and smarter. A new engine can lose all the legacy baggage. A new engine can adapt to new ways of doing or thinking faster and more completely. Look, for example, at the time and effort it's taking for the huge Unity team to produce their Data Oriented Technology Stack. A new engine designed from the ground up with a team a tiny fraction of Unity's size could be finished already without being constrained to getting the new ideas working alongside the old.

Thankfully the rest of the world isn't so quick to throw in the towel as you!

 

Honestly, I agree.

Wow that's an even more bold statement than I'm making.

Well, "next gen" (current gen now) started with UE5. There's no sign of it starting anywhere else.

I don't want us to get to the end of this gen and non-UE5 games all pale in comparison with UE5 games. But "table-top" demo videos are not showing scalable performant game engines - it's why it takes years (or an entire generation!) before demo visual techniques get into games.

UE5's ethic is to solve the scale of a game's entire render complexity, not simply to add nicer looking (physically plausible looking) GI.

So it boils down to what devs/studios/publishers think is the payoff for not using UE5. A lot of people seem pretty happy with last gen visuals... [Thread spawn.]

 

What are the UE5 games that you're talking about ?

 

A lot of people are still happy with PS2 era visuals. That's not what drives the industry forward though.
If the reasoning is that UE5 doesn't scale down well enough to be usable for 60 fps games on current gen systems then a) there's still UE4 for that - which supports a lot of RT stuff where necessary - and b) my expectation is that we'll see less and less of those games as soon as game devs will dump PS4/XBO h/w anyway so most cutting edge engines will likely end up at a similar performance profile as UE5 in a couple of years from now, with similar features and visuals.

 

The games that are coming before the current gen ends, whatever they are...



An alpha build of the game converted to use UE5, showing Lumen in a large world setting running on RTX 2080.

Never heard of this game. Comments in this thread:

https://forums.mmorpg.com/discussio...s-off-migration-to-unreal-engine-5-mmorpg-com

indicate a lot of low expectations about the game ever appearing. Small team, years and years of development, painful experience for some it seems...

 

TAA explained https://www.elopezr.com/temporal-aa-and-the-quest-for-the-holy-trail/

 

This paper discussing culling meshlets for skinned meshes seems promising.
"Conservative Meshlet Bounds for Robust Culling of Skinned Meshes"
http://diglib.eg.org/handle/10.2312/2633093

 

GDC presentation from last year on Red Dead 2's horses.

 

Hope this is the appropriate thread.