IbizaPocholo
NeoGAFs Kent Brockman
This is the kind of thing I want to see in the games, rather than more push to the graphics department. I want the improvement in physics.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
FLUIDNINJA Physics Plugin for Unreal Engine Looks Really Cool
- Thread starter IbizaPocholo
- Start date
- Opinion Trailer Platform
This is the kind of thing I want to see in the games, rather than more push to the graphics department. I want the improvement in physics.
Interfectum
Member
MiguelItUp
Member
Seeing this stuff is always really cool. Can't wait to actively see new and more impressive physics in games.
eyesabitdull
Banned
Man, a AAA god sim game sounds really nice with these physics engine.
Cockles & Whelks
Banned
3liteDragon
Member
Kupfer
Member
Tech is fine and all but now USE IT.
Last game I've played in which water effects really had an impact was From Dust, that was 10 years ago. Then, another 10 years ago, it was Blood Wake.
Devs have to figure out how to actually use water in an effective way. Just looking good and shiny isn't enough for me. TLOU2 water looks fucking awesome, but can I play around with it, fill buckets etc?
Last game I've played in which water effects really had an impact was From Dust, that was 10 years ago. Then, another 10 years ago, it was Blood Wake.
Devs have to figure out how to actually use water in an effective way. Just looking good and shiny isn't enough for me. TLOU2 water looks fucking awesome, but can I play around with it, fill buckets etc?
Ballthyrm
Member
It's always shown in big empty place with nothing around and not a lot going on.
Why ?
Because it's a huge ressource hog, it takes forever so make something that look that nice and then it runs only when everything else is absent.
The cost benefit analysis is not worth it for most devs.
It all about where you choose to put your graphical budget and your programmer/artist time and effort.
Not happening. The new consoles aren't nearly powerful enough.
Last edited:
ZywyPL
Banned
3liteDragon
Member
They’re powerful enough for both.
fran.fernandez
Member
Those volume texture simulations are around 10 ms. Is usable even in mobile….. but that is, don’t try to add anything more/
Hollowpoint5557
A Fucking Idiot
That smoke tornado was the tits.
Elektro Demon
Banned
Fuck yeah. I don't care for graphical advancements but the industry has actually regressed in terms of physics. It's sad to see modern games have zero physics in them, everything is pretty much flat and glued to the ground.
But if you take a look at some older games, max payne, fear etc, you'd see those games had solid physics in them.
But if you take a look at some older games, max payne, fear etc, you'd see those games had solid physics in them.
"Scalability". Cheap and boring scalability
GymWolf
Member
IbizaPocholo
NeoGAFs Kent Brockman
FluidNinjaLive.zip
drive.google.com
Dropbox - File Deleted - Simplify your life
www.dropbox.com
HouseDestructionDemo.zip
drive.google.com
Last edited:
IbizaPocholo
NeoGAFs Kent Brockman
Last edited:
SlimeGooGoo
Banned
.
Last edited:
Zannegan
Member
You can see in the video that it's all visual at this point (which, of course, because it's a tech demo). I just wonder how easy it would be to make those waves affect the character's movement or make the fire light, burn, and spread. Love the way the grass moves though.
Last edited:
Holammer
Member
Few games use the crazy physics they showed off with old Phys-X demos, or even ones shown after Nvidia bought them.
Like raytracing there's an uncomfortable computing cost associated with it.
BARRAD
Neo Member
any way to change resolution in the demo?
Shifty
Member
Looks pretty, though UE has had simulation tech demos on this level for years in the form of Nvidia's GameWorks stuff, and nobody uses that either because it's designed for top-tier PC hardware instead of the lowest common denominator.
I had to spend weeks talking our lead artist down from Nvidia's customized engine ledge back when I was doing VR work. It's seriously compelling to look at, but good luck finding a customer base with those hardware requirements.
These dense high-fidelity 3D simulations need to be computed wholly on the GPU to scale with interactive levels of performance, meaning their results are only available on the GPU unless you read them back into CPU land to affect gameplay objects.
That process tends to be performance-expensive, since it involves syncing up both units while memory is transferred. It can be accelerated with specially designed shared memory, or made asynchronous over a few frames like Metal Gear Rising's model cutting routine, but one way or other isn't a simple 'just works' affair.
It's probably more feasible to run a smaller-in-scope version of the same sim in directly in CPU land, since all that data would be available every frame with no fiddling.
Trouble there is that CPUs (the weak-ass console ones in particular) don't multithread nearly as well as GPUs, so you'd be trading off density or total size that way.
I had to spend weeks talking our lead artist down from Nvidia's customized engine ledge back when I was doing VR work. It's seriously compelling to look at, but good luck finding a customer base with those hardware requirements.
Not impossible, but certainly difficult to do right.
These dense high-fidelity 3D simulations need to be computed wholly on the GPU to scale with interactive levels of performance, meaning their results are only available on the GPU unless you read them back into CPU land to affect gameplay objects.
That process tends to be performance-expensive, since it involves syncing up both units while memory is transferred. It can be accelerated with specially designed shared memory, or made asynchronous over a few frames like Metal Gear Rising's model cutting routine, but one way or other isn't a simple 'just works' affair.
It's probably more feasible to run a smaller-in-scope version of the same sim in directly in CPU land, since all that data would be available every frame with no fiddling.
Trouble there is that CPUs (the weak-ass console ones in particular) don't multithread nearly as well as GPUs, so you'd be trading off density or total size that way.
Last edited:
Shifty
Member
Further to my point about it being more feasible to run these things in CPU land if you want them to affect gameplay, check this out:
Being a full voxel world rather than a special effects isosurface generated from a particle point cloud, it's not quite as visually impressive as the UE stuff in terms of sheer image quality.
However, since it's all done with a cutting-edge CPU algorithm, the entire sim can affect gameplay by default.
So in theory, you could take the sim data from this, then feed it into the GPU to generate pretty volumetrics via marching cubes, distance field raymarching, or various other techniques to get the best of both worlds.
The numbers would suggest being able to pull 120FPS on 4 cores, though unfortunately there's no mention of what kind of CPU it's running on.
Being a full voxel world rather than a special effects isosurface generated from a particle point cloud, it's not quite as visually impressive as the UE stuff in terms of sheer image quality.
However, since it's all done with a cutting-edge CPU algorithm, the entire sim can affect gameplay by default.
So in theory, you could take the sim data from this, then feed it into the GPU to generate pretty volumetrics via marching cubes, distance field raymarching, or various other techniques to get the best of both worlds.
The numbers would suggest being able to pull 120FPS on 4 cores, though unfortunately there's no mention of what kind of CPU it's running on.
Last edited:
SirKicksalot
Member
Flashback to 2009 when Carmack demoed a voxel-based id tech 6:
Neue id Software-Engine mit Voxel-Grafik - Ausblick auf die übernächste Grafikgeneration
Das Team rund um die Programmierer-Legende John Carmack arbeitet schon jetzt am Nachfolger für die id-Tech-5-Engine des kommenden Titels Rage. Das...
www.gamestar.de
Shifty
Member
I miss the Pre-Oculus era when Carmack would rock up with crazy new tech every few years. Megatextures may have been a miss, but those left-field innovations of his were always fascinating to see.Flashback to 2009 when Carmack demoed a voxel-based id tech 6:
Neue id Software-Engine mit Voxel-Grafik - Ausblick auf die übernächste Grafikgeneration
Das Team rund um die Programmierer-Legende John Carmack arbeitet schon jetzt am Nachfolger für die id-Tech-5-Engine des kommenden Titels Rage. Das...
Yeah, that was very impressive at the time. If memory serves it was in specific areas like narrow inter-zone corridors where it could look good without tanking the framerate.
Last edited:
kiphalfton
Member
Aww yes, the age old question "But can I play with it/them".
The answer...
You can look, but not glean any real enjoyment from it.
Last edited:
SirKicksalot
Member
Hellgate London and especially Stalker Clear Sky had much better versions in DX10. They did not require Nvidia or Ageia cards.
amigastar
Member
Never knew that Stalker and Hellgate had this feature. Gonna look it up on youtube.
IbizaPocholo
NeoGAFs Kent Brockman
IbizaPocholo
NeoGAFs Kent Brockman
CamHostage
Member
So, yes to UE5, but just FYI, FluidNinja physics doesnt have much to do with UE5. It can be used in development of UE4 projects just the same.
UE5 is going to be great when projects built for its advancements start to finally ship, but we should be a little careful to think of it as the "savior of next-gen"; it's a tool for harnessing the full power of these machines (and other machines, existent and to come) and advancements in simulation systems, and there are lots of these tools in the works or being employed in upcoming software. UE5 is not the thing; it's the thing that helps us get to the thing.
Last edited: