/v/ - Video Games

Who would have thought 3DPD also applied to gaymes?

>>14553591

So is virtualizing geometry. This was actually the plan Carmack had for id Tech 6 before he left. The future won't be fast parallel triangle calculators.

>>14553623

Virtual Texturing aka Megatextures is technology thats actually possible in consumer gpus thou

as Doom and Wolfenstein proved running ID TECH 6

I would want to see Ray Tracing that has slightly higher requirements not something ridiculous thou

>>14553681

>Virtual Texturing aka Megatextures is technology thats actually possible in consumer gpus thou

that was in id Tech 5, they cut the virtual geometry and voxels from 6 (crytech guy took over)

>>14553696

>crytech

They still alive?

>>14553607

even amds low end IGPs are running games just fine. You don't need a 1080ti, and it's not like there's any games that really take advantage of it. We need another HD4800 series to shake things up, something not good for miners but solid for people who just want to play games that undercuts the competition significantly in its price range and gets similar performance to hardware that's double or triple the price tag.

>>14553728

in terms of AI they mean actual learning AI and not NPCs in video games finding their way through a doorway.

>>14553710

I don't know if CryTech is still alive, but I know that one of the main programmers from CryTech is now in charge of id Tech 6 since Carmack left.

>In 2008 and while id Tech 5 had yet to be fully formed, John Carmack said the next engine by id Software would be looking towards a direction where ray tracing and classic raster graphics would be mixed.[3] The engine would work by raycasting the geometry represented by voxels (instead of triangles) stored in an octree.[4] Carmack claimed that this format would also be a more efficient way to store the 2D data as well as the 3D geometry data, because of not having packing and bordering issues.[3] The goal of the engine would be to virtualize geometry the same way that id Tech 5 virtualized textures.[5][6] This would be a change from past engines which for the most part use mesh-based systems. However, he also explained during QuakeCon 08, that the hardware that would be capable of id Tech 6 did not yet exist at the time.[7]

>In July 2011, Carmack explained that id Software was beginning research for the development of id Tech 6.[8] It's unknown if Carmack's vision of the engine at the time was still the same he described in 2008.

Then…

>Following Carmack's departure from id Software, Tiago Sousa, who had worked as the lead R&D graphics engineer of several versions of the CryEngine at Crytek, was hired to lead development of the rendering.[17][18] Bethesda's Pete Hines has commented that while id Tech 6 reuses code written by Carmack, most of the decisions made about the engine's direction were taken after he left.[13]

>>14553753

They're already kind of onto that with things like the Ryzen 2200G. I think it needs a bit more power behind it and to be a discrete GPU in the 200-300 range that doesn't need to get 1080 performance but maybe something like Vega32 or whatever if it ever comes out will scratch that itch.

WHERE'S THE FUCKING VOLTA GAMING CARD JENSEN.

>>14553772

Carmack was a god tier coder shame he fucked off.

>>14553778

that's the virtual textures. there is none of the raycasted voxels aka virtual geometry.

>>14553780

>Carmack was a god tier coder shame he fucked off.

Yeah, but I suspect he can foresee the limitations of the hardware at least 10 years down the road… so I'm really hoping his research and work into "functional" VR pays off better than game engine designing.

>>14553793

>"functional" VR pays off

I legit don't see VR gaming ever being a thing, for engineering and other uses I can see it being huge.

>>14553808

>I legit don't see VR gaming ever being a thing, for engineering and other uses I can see it being huge.

Most of what I recall reading from interviews with Carmack about this is how much streamlining and optimization it requires. I'm pulling this number out of my ass, but he said the feedback needs to be down to a few milliseconds for most people. Otherwise your brain "desyncs" from your equilibrium and you run the gambit from being nauseated to feeling "lagged" from the VR environment.

So the reason all VR graphics are shit is because more visual processing requires more time which increases the feedback loop timing which increases the severity of the "symptoms".

>>14553820

I actually think he was talking about ditching meshes and polygons altogether. Since those are completely irrelevant to raytracing.

>>14553793

Pretty sure Carmack was talking about making Voxel Octrees, the data format

worked years to make it work as geometry, it didnt pan out or whatever (who knows what happened there)

and later moved to VR instead

>>14553838

>it didnt pan out or whatever

He literally came up with the idea almost 10 years ago and at the time said the hardware required to make the move didn't exist yet.

But anyway, the octree would be to optimize storage and processing for both sets of data: the virtual textures and the virtual geometry.

>>14553591

Isn't one of the reasons as to why Project Offset was cancelled is because of the failure to implement real-time ray tracing?

>>14553884

I think it's actually extremely difficult to fail making a real-time ray tracing renderer. It's just that for last 2-3 decades it has always been way too computationally expensive. That's for sure one of the reasons Carmack said back in 2008 the hardware didn't exist yet.

Intel's Knights Ferry or something based off that R&D was looking very promising for real time ray tracing, but they spun it into a non-graphics processor addon or something I forget.

>>14553910

>These massive computers being built at an unprecedented rate are the reason for the high price of GPUs and DDR4 memory

That's actually a load of bullshit that everyone who has passed econ 101 can see plainly. You might as well claim that IBM's z13/z14 are the reason Intel i-series costs are higher too.

>>14553884

proprietary fixed function GPGPU-CPU hybrid hardware (not a GPU!) that Intel thought could compete against the entire graphics industry

circular internal politics and delays with the complicated specifications killed it after it no longer was viable as regular consumer product

I hope this Ray Tracing tech wont make the same mistakes

>>14553967

Ironically it already has.

Whatever the future of Intel's Knights Ferry holds, it certainly still has real time ray tracing in its grasp above all other "competitors".

>>14554172

I know these are pathtracers, but still interesting. I don't really know the difference between this and raytracing.

>>14553591

this shit won't make your games any less shittier. If anything it'll now bloat development time as they will now have to adopt CGI levels of scrutiny and effort which will in turn only blow out budgets even further. And no ones gonna be able to run this shit anyway without the best hardware.

Graphics fags please fuck off.

>gaytracers

fun fact: graphicsfags are all soyboys from san francisco

real men play text adventures

>>14554602

Is that an e-paper screen? I was actually thinking how good of an idea that might be to have that for reading and typing shit so you don't go blind looking at a normal screen all the time.

>>14554646

It is, it's not a bad idea but they absolutely refuse to add line editing and it's waaaay over priced.

Has e-reader internals. Could be a couple hundred bucks max.

>>14554646

It is. It's also five hundred fucking dollars.

In and of itself, having a simple no-frills word processor could be a useful thing. The freewrite is not a bad idea, but at that price, it's fucking ludicrous. You can get a 6-inch e-ink reader for $60. You can get a mechanical keyboard for $80. You can get a SOC computer (like a Raspberry Pi) for $20. This thing is nothing more than components from those things in a metal housing, for $500. If you made one of these things and sold it for $100 to $150, I'd honestly get it. It wouldn't be my bag and I wouldn't get it, but I'd understand. It's an electronic typewriter (though not being able to go back and edit is fucking stupid; you have to actually delete back completely to the spot you want to change), and the draw to a no-distraction electronic typewriter should be obvious to anybody who enjoys writing.

What it is is fine, but there is absolutely no excuse for the price. You'd have to be a hipster retard to pay that much for something like that.

Is blender the only one that can do ray tracing right?

>>14554727

Blender doesn't "do" raytracing. Blender's internal renderer isn't even a raytracer, it's a rasterizer. Blender otherwise defers its rendering to external renderers, though the current canonical raytracer, Cycles, is actually developed by the Blender project

Blender's raytracing also isn't real-time. You can't use the same techniques that Blender uses for real-time raytracing, because it's designed around high-quality photorealistic effects rather than high-performance raytracing.

>>14553910

Actually, that is only a dent compared to crypto.

>>14553623

Well please give me the redpill on what virtualizing geometry is about, anon-kun

>>14555287

the dragon boss fight/monk fights were part of a demo reel one of the animators used for his portfolio. Shame it appears to have been lost.

>>14555352

Never ever for consumers. You gaymerz are getting Turing, which is just a tweaked pascal. Enjoy your Nvidia goyworks on a GPU named after a literal faggot.

>>14559605

We must secure the existence of our people and a future for white children.

>>14555185

>Well please give me the redpill on what virtualizing geometry is about, anon-kun

You ditch polys/meshes and raycast voxels.

This:

https://en.wikipedia.org/wiki/Ray_casting

plus this:

https://en.wikipedia.org/wiki/Voxel

plus probably this:

https://en.wikipedia.org/wiki/Volume_rendering

>>14559605

>goat tower

>bug

>power line

>vegan pizza

>bug

>blezsinski windmill

>kamen rider

>space elevator

>clay statue of disapproval

>animu

>flower

>bug

>your mother

>nippon steel, but chinese

???

>>14559605

Spiral staircase

Rhinoceros beetle

Ruins street

Fig tart

Rhinoceros beetle

Singularity point

Giotto

Angel

Hydrangea

Rhinoceros beetle

Singularity point

Secret emperor

GET FUCKED, NOW I RESET THE UNIVERSE!

>>14553926

>>Only thanks to Amazon buying their engine.

>Are you serious?!

google Amazon Lumberyard

>>14555411

Source to FUCK MY SHIT UP,

>>14554375

>If anything it'll now bloat development time

Actually Raytracing has the potential to shorten development time as it cuts out a lot of the work arounds and tricks that are nowadays necessary to do (ex. cubemap reflections)

Implementing raytracing means that reflective surfaces will accuratelly reflect their surroundings, so no need for all that bs. Makes texture artists job easier as well and also for the guy who sets up lights and cameras in the scene

>>14560526

That would be true if the programmers are competent.

>>14553728

Which game is that stage left?

>>14554602

>low set ears

>dark eyes

>DAT NOSE

that's a yid.

>>14559179

>>14559198

reminds me of that time some (((media outlet))) was shilling fried bugs as a food source because muh environment and that it's selfish to eat meat

>>14561141

Competent people cost money and expect to be treated with some measure of respect. The people who run video game companies want slaves they can pay pennies and beat down for fun. Pajeet will work for pennies and is used to people above him treating him like vermin.

>>14553591

that entire scene was using raytraceing while all the other tech demos used a mix of rasterization and raytracing.

>>14555352

well its taken for granted that the next cards will be using GDDR6, SK Hynix will be in mass production starting in 3 months.

so sometime this year so long as its atleast about 4 months from now.

>>14561141

>That would be true if the programmers are competent.

ray tracing is easier to program than rasterizing

Gaytracing won't even fucking matter unless devs actually utilize it in a meaningful way

AKA any "cutting-edge" graphics technology

>>14554602

Does it have a printer though?

>>14561152

bioshock infinite

>>14553753

>The older games feel more next gen then half of the shit now.

>Red Faction: Guerilla still has the best destruction physics engine of any game I've played

>STALKER still has the most interesting AI I've ever seen in open world games

>Half-Life 2 still has better and more functional physics than a lot of games that came out just last year

>F-Zero GX still has the greatest sense of speed out of any racing game

I blame the increasing cost and time of AAA development. No one can take risks anymore because AAA take of the cost and time being so much that your game needs to be a runaway success or else your studio will be shutdown.

I hear a lot of the upcoming rendering tech will make development easier but I don't actively research or follow any of that stuff so I don't know. I hope it is though, because easier development means less expensive and less time consuming, which means more polished and content richer games.

>>14561514

Don't embarrass yourself. Raytracing, by design, is capable of providing photorealistic grafics. It's been around forever and it's used in rendering. The problem is that we haven't had the computing power to use it in real time until JUST now.

>>14559865

Jesus fuck what a fucking fate.

>>14561612

The production of most AAA western games are super bloated.

>>14561637

Even with the bloat it doesn't change the fact that costs have been going up and the time it takes to produce art assets has been going through the fucking room. To the point where the FFVII remake can't possibly have everything the original did because development of art assets simply takes too long. Any new rendering tech that can alleviate some of the art development headaches would be a blessing upon this industry.

>>14561657

You can have all the tech you want it doesn't mean shit if you can't make a decent game.

>>14561665

This is so true.

>>14561612

>easier development means less expensive and less time consuming, which means more polished and content richer games.

I wish I shared your optimism. Easier means developers can be lazier, and producers demand even more stringent time constraints because they expect the same (shit) quality to be produced in faster time.

>>14561624

>The problem is that we haven't had the computing power to use it in real time until JUST now.

That's not entirely true, either. Real time raytracing has been shown off in the demoscene for ~20 years now. The problem is it used up basically all compute cycles leaving nothing for gaming logic.

>inb4 PhysX 2.0 where you have to buy an extra addon card to get it to work properly.

>>14561327

>that's a yid

That is correct.

Bonus Points if you also thought that it is not his idea but that he ripped of/bought for pennies a students masters project and refuses/is not able to alter the features.

>>14561521

It can only upload to the Freewrite cloud service from where you can then export it to evernote etc to download the txt file.

There is no way to get the text without the cloud service.

If Freewrite shuts down or you have no WiFi and Internet connection you have bought an expensive paperweight.

>>14567615

I've had the exact opposite experience, anytime I say AMD is shit at OGL so you don't want one for emulation there's always one or two reply that are one of those

>"Works for me™"

>"muh shills"

>"but it's already been fixed, AMD says so"

>>14567556

You're obviously way too young to even know about math co-processors like the 287 and 387.

>>14567729

Not exactlywhat I'm referring to, PhysX unless you go back to the Ageïa days doesn't require special hardware, but anything that actually uses PhysX well needs you to have a second GPU or suffer shit framerate (Arkham series and Alice Madness Returns being very good examples of that), I expect about the same shit will happen with raytracing.

>>14567710

I never saw it, so I'd need evidence but it's generally accepted that AMDs openGL support isn't very good. Not like OGL matters going forward, but for legacy support it does. Either way, it's clear as day PC's corporate cock suckers are the most annoying.

>>14567710

I just think its weird that AMD is always brought up even when it's just talking about some problem that is exclusively nvidia related and has nothing to do with AMD.

>>14567743

It's exactly what you were referring to and why I said you're obviously way too young. Floating point math doesn't require special hardware either. Intel still made coprocessors specifically to accelerate that math.

>>14567766

no it is, but whether it's legitimate shills or fanboy faggots, it's hard to tell. Nvidia has the money but there's a certain breed of fanaticism in PC hardware that is unlike any other.

>>14567770

>>It's exactly what you were referring to and why I said you're obviously way too young

Was reffering to PhysX which can be both accelerated by specific hardware and generic GPUs

>. Floating point math doesn't require special hardware either

And yet that floating point coprocessor can't do anything else but that, so it is special hardware

Stop trying to look smart when you can't even read my post.

>>14553591

Wasn't raytracing the future when the GTX 400 series came out?

>>14567815

It's been a long time coming.

http://www.q3rt.de/

>>14554182

Raytracing shoots rays out of the camera per pixel. Pathtracing shoots a ton of rays out of lights, has them bounce around hitting stuff and reporting that back to the camera.

In effect this means path tracing is more physically accurate, has actual area lights, and scales better at higher resolutions, but the downside is that you usually need more rays and don't always get a result per pixel, so the image has to converge over time. The latter is what gets solved with AI, which is trained to use the previous frames and what it's learned about rendering to turn a single noisy frame into what looks like a converged image considerably faster than it would take to actually converge (which can be hours). Downside to that is that it gives less physically accurate results.

>>14567793

nice back pedal:

>Was reffering to PhysX which can be both accelerated by specific hardware and generic GPUs

PhysX is literally analogous to what I said. Floating point math doesn't need specific hardware just like the PhysX engine doesn't, and both can be accelerated with specialized hardware (pro-tip: GPUs are specialized hardware).

>Stop trying to look smart when you can't even read my post.

kek'd because you actually got rekt

>>14567615

Because both piss me off with their bullshit but for different reasons, people tend to ignore the bullshit AMD pulls due to Nivdia being so out in the open scummy.

>>14568829

>nice back pedal:

Can't be a backpedal if it's the first thing I said.

>doesn't need specific hardware just like the PhysX engine doesn't

Actually it does now that I think about it but someone as smart as you should already know that.

>pro-tip: GPUs are specialized hardware

Just for the sake of semantics, I meant specialized as in single task not specialized as anything that isn't non-specialized.

>kek'd because you actually got rekt

>come out with a post that has only tangential relevance to what's being said

>declare it an argument

>act like you won said argument

t.pigeon playing chess

Or are you gonna pretend that what you posted has any relevance to me pointing out that raytracing is gonna be used as yet another excuse to sell us hardware specifically for it thus making the whole thing mostly moot since only a few games are gonna do it and any specific implementation is either gonna be non-fonctional or extremely gimped on part of the hardware.

>>14553591

i dont want to defend nvidia

but you know that this shit is not for gaming right?

this is their "milk the shit out of science comunity" product line

>>14569054

>Wrong. PhysX is a software engine that can be accelerated with specialized hardware.

But PhysX does need specific hardware in practice, although specific in the more general non-computer related sense of the term.

>GPUs are the modern day equivalent of of the x87 FPUs; and the PPUs in these modern GPUs are too. Still, the fact that both floating point math and the PhysX engine don't require any specialized hardware makes inserting that argument moot.

Your argumentation is more and more bizarre, it's almost like you're incapable of seeing that I already said PhysX doesn't need specialized (as in single task) hardware but that but you could use some anyways.

>I don't need to pretend.

No you clearly don't

>you even got it wrong trying to claim PhysX isn't software engine capable of receiving hardware acceleration

But I never claimed that, yet again you miss the point because you can't read beyond what triggers your autism.

It's incredible how much of your post is repeating things I already said but disagreed with and things that are true but never had relevance in the argument outside of your head.

>>14569312

>You are incapable of understanding that one can run floating point calculations, PhysX, and OpenGL without any FPUs, PPUs/GPGPUs, GPUs; all in the CPU

Or you know maybe that's because you're incapable of seeing that I already said that.

>I did you the courtesy of trying to explain how and why you are wrong, and you disagreed.

You explained things that had no bearing on the point I tried to make so I tried to reiterate that my point wasn't purely theoretical but what actually happened in the real world then you went on a storm of autism while still being incapable to understand what I'm writing.

Now to try and actually be polite once again my point in the first two posts was that people are gonna use raytracing as a way to fuck consumer over in every way, like PhysX, are you capable of understanding that?

It doesn't matter that it is similar to something that was done in the past, if anything it points out that we're still getting fucked over in the same way but now it's even worse.

It doesn't matter that there's no technical barrier to running that on any hardware because manufacturer will put those restrictions in place anyways like what happened with PhysX and that very likely is what will happen with the multiple proprietary implementation of raytracing.

And it certainly doesn't fucking matter that CPUs could do all that by themselves at some point in the future because that is unlikely to happen for the foreseeable future unlike aytracing being implemented in consumer software.

You are completely incapable of seeing why all your arguing is completely besides the point I was trying to make and the saddest part is that you're trying to be right in an argument where I didn't disagree.

>>14569382

>proprietary implementation of raytracing

I told you stop talking about things you don't know anything about.

>And it certainly doesn't fucking matter that CPUs could do all that by themselves at some point in the future because that is unlikely to happen for the foreseeable future unlike aytracing being implemented in consumer software.

Yeah just like x87.

You're a waste of space. Kill yourself.

>>14569986

>I told you stop talking about things you don't know anything about.

Please do enlighten us as to how DXRT and RTX aren't gonna be proprietary.

>>14570063

>Please do enlighten us as to how DXRT and RTX aren't gonna be proprietary.

Please do enlighten us as to how real time raytracing requires the use of proprietary anything.

Here's more real time raytracing (this time only 18 years ago) done in just 64k. The fact you won't shut the fuck up about shit you literally have no experience with, and obviously no knowledge about is now screaming FUD.

Real time raytracing will be a CPU thing. It will never be a GPGPU thing despite faggots like you spreading your stupid FUD.

>>14570135

Real time raytracing from 2009 done in under 4k.

>>14570135

>Please do enlighten us as to how real time raytracing requires the use of proprietary anything.

It doesn't per say, but if you seriously think there's not gonna be a massive push from Nvidia or M$ to use their own implementations of it which will no doubt include artificial restrictions you're naive.

If you're gonna deny that proprietary APIs even exist because there's an open one then I don't even know what to say.

>It will never be a GPGPU thing despite faggots like you spreading your stupid FUD.

Metro Exodus already exists so yeah it's already a thing, but I guess denying reality is your thing.

>>14569054

>Also when AMD does their bullshit I've tended to feel less fucked over than when nVidia does; because at least I saved a bit of money with AMD.

I'm not forgetting the shit AMD has done but I'm pleased with how good Ryzen is so far.

More 4k realtime raytracing.

>>14570173

>If you're gonna deny that proprietary APIs even exist because there's an open one then I don't even know what to say.

I am not denying one exists now; I'm just denying it means what you think it means. I personally will be pleased with a standardized API because I actually do this for my job and my hobby.

I'm just denying your FUD about the future hardware. Every video I've embedded so far was done without an API (proprietary or not), and without specialized hardware.

>>14553937

if youre arguing that it is actually miners that are driving the prices up ur dum

if youre saying neither this nor supercomputer production are driving prices up you might not be dum but id like some more evidence than

<supply and demand

>>14570178

>I'm not forgetting the shit AMD

Honestly I can't recall many really bad thing AMD has done outside of those:

>heavily marking up the price of some CPUs back when they had the best performance (FX-57 and friends that you could beat easily with a cheapo Opteron or 4000+/4800x2)

>Phenom launch with TLB bug

>Bulldozer launch with maximum damage control and faked/biased leaked benches

>cutting legacy support entirely around 2016

>RX 480 launch and fried mobos

>encouraging the buttcoin mining craze twice

>>14570204

>I'm just denying it means what you think it means.

So you're denying that DXRT is gonna be windows exclusive until someone puts out a wrapper for it and that RTX is gonna be gimped or non-functional on non Nvidia hardware, safe bet to take that you're wrong.

Hopefully less shit APIs become prevalent like currently with Vulkan, not holding my breath though, since the first commercialk game with real time raytracing is gonna be DX12 and Nvidia exclusive most likely.

>I'm just denying your FUD about the future hardware. Every video I've embedded so far was done without an API (proprietary or not), and without specialized hardware.

Everything you've shown so far is a demo with likely completely custom solutions that only apply to their own scope, it's completely unreasonable to think they mean much outside of the scope of the demoscene which has always done shit that is completely insane but never really reflected elsewhere.

How about you post a practical demo of an actual game, you know like that Q3 one from 2006.

>>14570228

they also did some image quality botching way back in the early 2000s in games like quake 3, making it look significantly worse regardless of settings in order to get higher frame rates. AMD/ATi have done a lot less worse than the competition, and the retards who cry "buh wut about AMD" when nvidia/intel flaws are brought up need to shut the fuck up

>>14561327

>reminds me of that time some (((media outlet))) was shilling fried bugs as a food source because muh environment and that it's selfish to eat meat

I wouldn't eat fried food, but there is Chapul protein, which isn't half bad.

>>14570228

Nivdia are total shit stains who need a bullet and I'm not trying to downplay the shit by bringing up the AMD stuff.

>>14570228

The two times AMD cards have been massively sold for mining, it was not possible to directly mine Bitcoin profitably on GPUs. It was Litecoin mining (and other Scrypt coins) the first time around, and Ethereum mining this time around.

I did however mine Scrypt coins on my 280X cards and then trade it for Bitcoin, it worked quite well.