At 3 min he was driving the car in first person and the shadows falling on the car looked wrong. There was no penumbra.
Unreal Engine 5, [UE5 Developer Availability 2022-04-05]
- Thread starter mpg1
- Start date
cheapchips
Veteran
Satisfactory's latest update is in testing, where they're making the switch to UE5 from 4. I thought it was an interesting case of a game in transition. Since they're open world with a many objects, WPS / on file per object are big wins for them. At present they have to load in entire areas as players move through the world.
They've had to redo all their vehicle physics for Chaos.
On the graphics side, it's an interesting mix of features. They're not using VMS. They support Lumen, but since they have to support lower end cards it's a toggle on feature. They're not relighting the game for it.
They're using Nanite for cliffs etc, but not for objects yet. I'd imagine they'll benefit from it on factory pieces down the line.
They've had to redo all their vehicle physics for Chaos.
On the graphics side, it's an interesting mix of features. They're not using VMS. They support Lumen, but since they have to support lower end cards it's a toggle on feature. They're not relighting the game for it.
They're using Nanite for cliffs etc, but not for objects yet. I'd imagine they'll benefit from it on factory pieces down the line.
I wonder why -- wouldn't dumping everything into nanite result in much better culling and performance? Maybe just not a technically experienced studio, it's weird that they waited on an engine update to implement a file/world streaming system which suited their needs.
cheapchips
Veteran
On streaming, pobably more a case of working with/around UE4s limits and now there's something better. It's not like Satisfactory lacks technical scope. The factories you can build are nuts.
The hesitancy on going full Nanite adoption is probably just a case of one thing at a time. Factory objects seem like a good fit for it, but they're working LOD miracles already. It'd imagine there's a fair amount to unpick.
Inuhanyou
Veteran
There are not that many AAA studios anymore in general is the sober answer. The devs that have already been using ue4 will move over the ue5. In addition many of the rest are more comfortable using their own engines they have been using already. So it's not really that surprising
DavidGraham
Veteran
Let's see.
The Foundation engine for Tomb Raider was replaced with UE5.
Red engine for The Witcher was replaced with UE5.
Frostbite engine for Mass Effect was replaced with UE5.
X Ray engine for STALKER was replaced with UE5.
The custom engine for Silent Hill was replaced with UE5.
That's it, if you really look into it, 3 of these engines were replaced because they are too old, or too clumsy (Silent Hill, X Ray, Frostbite). Red and Foundation are semi old too, but there are probably reasons other than that for why they got replaced.
@DavidGraham The next Halo game is supposed to be Unreal Engine. Can't remember how legit that news was.
cheapchips
Veteran
It was a Bloomberg report from Jason Schreier that was then widely repeated.
Microsoft Studio Behind Halo Faces a Reboot on Years of Turmoil
The title has suffered multiple delays, development setbacks and staff changes
www.bloomberg.com
They're probably skeletal meshes or something, since they do animate. Pretty much all the animation is rigid and so it could probably be attached Nanite objects attached to bones (like the AncientGame demo) but that likely comes with its own overheads.
Seems reasonable to move the obvious static meshes first and see how that goes.
Nanite without VSM has some definite performance gotchas, but if they aren't making much use of shadowed local lights that should avoid most of them.
Last edited:
Shooty Fruity Unreal Engine 5 Update
Enjoy increased resolution and minor bug fixes to both the Quest and Quest 2 versions of Shooty Fruity!
ndreams.com
Did they release a 4k version of this video? The compression kills me!
Does not seem so. It's quite incomprehensible, really.
Very exciting times with ML which I'm sure will be transformative in workflows and combining many different fields will allow small team content creation, iteration times, scope etc to expand significantly. There'll be "base" models available eventually that'll be good enough for 90%+ of projects with that continually increasing as the training sets also improve for these animations. Said it before but imagine with synthesised voices improving (that'll be on the roadmap to implement you'd think) and matching them up with ML facial animations - very small games could be voice acted somewhat competently, relatively quickly and cheaply, which could be amazing for story and character heavy games. Sound effects too? Imagine writing prompts for all the sounds you could ever want and sliders adjusting things, or possibly exporting those sound files into
Getting ahead of myself here but we're still at the beginning of the next big S curve for many different things and hopefully that'll mean significant positive progress
Getting ahead of myself here but we're still at the beginning of the next big S curve for many different things and hopefully that'll mean significant positive progress
Last edited:
Inuhanyou
Veteran
I know some people are tribal about self made tech and in certain cases for good reason. But I am most excited about what devs focusing on ue5 can bring to the table. What devs of different talents can uniquely create from these tools that are only getting more varied and more flexible every time we see them
Similar threads
- Locked
