VR Megathread: Valve's VR Has Entered The Game

The Something Awful Forums > Discussion > Video Games > VR Megathread: Valve's VR Has Entered The Game

SCheeseman: Apr 23, 2003

They need to nail eye tracking before releasing v2. Rendering in high resolution only where needed would drastically reduce render loads while delivering a higher fidelity image. The tech would be worthwhile even on existing hardware if you supersampled in the FOV hotspot.

# ¿ Aug 29, 2016 20:50

Adbot: ADBOT LOVES YOU

# ¿ May 13, 2024 16:25

SCheeseman: Apr 23, 2003

Hadlock posted:

Foveated rendering sounds cool as poo poo, but I'll believe it when I see it. I don't see that being a v2 feature, and none of the Chinese knockoffs will have it. Sounds like a premium v3 feature at best.

Sounds like to me that foveated rendering will always be like SLI: improves performance on paper, but software support always lags or is buggy, and usually it's just easier to buy a single card of the next generation instead for the same performance, but without the added complexity.

Once we start getting to something like 4000x4000 per eye, foveated rendering will effectively become necessary. The amount of power wasted rendering pixels that aren't needed would get ridiculous. For the moment, it's only a bit wasteful.

SLI is completely different in that in order to do it well you need to communicate with the nvidia driver team since it relies so heavily on hacks in the driver, as well as a hell of a lot of work deep in the engine's code to make everything work running on 2 (or more) GPUs. It's a mess to implement and has been from the start, limiting its support to mostly AAA titles. Foveated rendering is an engine-level optimisation that doesn't require anything tricky like parallelisation, if Unreal and Unity support it then almost everything that uses those engines will.

# ¿ Aug 29, 2016 21:05

SCheeseman: Apr 23, 2003

The market is open to being fragmented by design, it's the PC market! Any company can make an OSVR/OpenVR headset and motion controls.

# ¿ Aug 29, 2016 21:10

SCheeseman: Apr 23, 2003

It kinda seems like it sucks

# ¿ Sep 1, 2016 14:38

SCheeseman: Apr 23, 2003

beergod posted:

So I'm trying to lay on my couch and watch a movie in Virtual Desktop with the Rift and having no luck.

When I look at the ceiling, it's just a black bar. Id like to refocus the screen I can watch the movie by staring at the ceiling, preferably in a way that requires no keyboard input after the movie starts so I can just lay down and watch it.

I've heard people suggest hitting F4 after the movie starts, but it's not refocusing it on the movie. I still have to move my neck weird to watch it.

How does everyone else do this?

https://steamcommunity.com/games/457550/announcements/detail/885342693256166654

# ¿ Sep 6, 2016 02:55

SCheeseman: Apr 23, 2003

The FOV difference between the two is negligible.

# ¿ Sep 9, 2016 15:55

SCheeseman: Apr 23, 2003

I love VR but unless you're ultra-excited about it and willing to beta test just about everything I'd say wait, maybe for Gen2.

# ¿ Sep 16, 2016 09:41

SCheeseman: Apr 23, 2003

Poetic Justice posted:

Holy poo poo, that's amazing. Good guy Insomniac

The game plays like a mediocre PS2-era game and apart from the VR stuff is presented like one as well. It sucks and they're acknowledging it sucks. I guess it's honest?

# ¿ Sep 17, 2016 04:46

SCheeseman: Apr 23, 2003

Poetic Justice posted:

Yeah I bought it, haven't played it yet, but I was talking more about giving away all those other games to people that bought Feral Rites at full price.

Which is nice, but they can only do that so many times. It's just baffling how the game even got made.

# ¿ Sep 17, 2016 07:33

SCheeseman: Apr 23, 2003

Different strokes. A warmed over relic from 10 years ago in VR is less interesting to me than something that actually tries to do something new with the medium, early access or not.

# ¿ Sep 17, 2016 07:49

SCheeseman: Apr 23, 2003

They have a lot of money to throw around.

# ¿ Sep 17, 2016 07:53

SCheeseman: Apr 23, 2003

Here's a good writeup from a VR developer (the dude behind the excellent and innovative Hot Dogs, Horseshoes and Hand Grenades) about challenges faced from VR development.

quote:

Hey folks, A little late to the party, but I thought I'd add my two cents to this conversation, and share a bit of the challenges that I've faced, and some observations I've drawn from tackling the problem space of scaling systems/spaces/etc. for VR games.

There are a number of challenges faced by devs moving to VR (big and small) that I don't think folks who don't do game dev are quite aware of the magnitude of. These relate to what are called implementation patterns, patterns used for things like character controllers, physics objects, spatial event triggers, visual effects, the list goes on. Games from AAA to one-person-indie are in a sense, always resting on, building on top of this shared body of arcane knowledge on 'how we do things'. All the little stuff that contributes to things in games that you don't notice when they work well, but DO notice when they don't:

How a character model walks across a bumpy surface, or up and down stairs.

How heavy vs. light objects interact with the player/other forces, when and how they come to rest

How a door works

Fire/explostions looking like fire/explosions

The ability to rapidly jump between gameplay and interface manipulation situationally

A Main menu

The list goes on. In many ways, the 'indie revolution' in gaming, and frankly the rising across the board software quality of AAA games has occurred because of the fact that successful patterns, in abstract and in code, have become significantly easier to share/spread over the past decade, and many of those problem spaces have been effectively 'solved', meaning that there are 1-3 ways to do a given thing, because its the most reliable/easiest to implement for a desired behavior/context, and everyone just does that.

Now imagine a majority of it has been thrown out the window. THAT is where we are at.

And not just for one reason/property of VR, for several of them. The most prominent that I've dealt with are:

Stereoscopic Rendering
Far more of the content in traditional games we play is flat than you'd imagine. One can occasionally get a glimpse of this glitching a camera into an effect, or if its just poorly built, but seriously think for a moment about how pretty much every visual effect you see in a game is a bunch of flat 'cards' overlapping each other. Most of these look immediately wrong/bad the moment you introduce stereo (this is also incidentally why vfx in post-converted 3d films often look so stupid). We have just gotten to the point where its performance efficient enough to have truly volumetric effects in regular games (fog, pyroclastics, simple liquids, etc.), but all of these effects get GEOMETRICALLY more expensive for larger resolutions/higher framerates. Which brings me to:

VR Resolution and Framerate
1080x1200x2x1.4xPrewarp Supersamplingx90fps = 326592000 pixels (assuming a single pass, no overdraw at all) This is more pixels than full 4k gaming at 30fps, but in a context where using motion blur is out of the question (as most AAA games do), and dropped frames result in physiological discomfort. Plus, due to having two views involved, despite some major recent advancements in the common engines, the cpu overhead is much higher than rendering single 4k frames.

How to put this... we have gotten used to, especially with AAA games, of eating frame stability in the name of content. Other than Order 1886's rock solid 'cinematic' 30fps, and the latest DOOM (which is a technical tour de force), I can't remember the last time I played a major release game that didn't oscillate WILDLY draw-time wise. Even shops like Blizzard that used to be way better about this feel like they've knocked this down a few notches on their priority list.

So we now have a situation where the way we make everything, measure and create content, compose scenes, and combine everything at the end is a gun to our head. AAA productions in particular are not ready for how game-changingly hard this is. Those big games like Assassins Creed and Watchdogs? Yeah, major scenes, assets, mechanics, etc. were integrated into the final build mere weeks before those games went to cert. That ain't going to fly making VR games. An entirely new philosophy needs to be adopted oriented on significantly more organized pre-production, technical iteration and perf. testing in cycles throughout productions, frankly... constraining content production in ways that these large edifices aren't used to having to do. There is no longer a 'well, it doesn't run perfectly, but it works well enough' grey area. You hit frame-time, or you don't.

When it comes to the first AAA VR titles, expect some of them to cock this up ROYALLY. DOOM VR will probably run amazingly, because ID locked that poo poo down. Fallout 4 VR will probably run unreliably. I wouldn't be surprised to hear that 90+% of their resources in making that happen are about perf.

Physics
This is one that will hit folk differently based upon what type of game is being made, that I've grappled with to a preposterous degree in H3VR, because Anton made the quixotic decision to go 'everything's a physics object'. But even if you're not going to this extreme, it's important to realize that game physics engines are not architected to handle what we're already asking of them for VR games. They handle coarse interactions between man-sized objects very well and very fast. They handle vehicles... pretty drat well, if you're willing to throw 2-3 NERDS at the problem for the entire dev cycle of your game, and especially if you're building off a code-base that's been refined over a series of games (which is why we have so rarely new break-out IPs in those areas).

What they DONT handle well, is things the size of your fingers, or even your hands. Or the size of most of the objects you hold in your day to day existence. Keys and locks, glasses, mice, realistically sized levers/handles/etc. Even something like... using a crow-bar to twist-break a chain off. Utterly outside the realm of said engines to do without it being mostly non-physically canned. This is why thus far, most folk have gone with very arcadey, very loose/forgiving, non-physics-based solutions for common interactions that.. could be physics based (I almost said 'should', but that's just my position on things, certainly not a board truth).

One of the core value propositions of VR is this ephemeral notion of 'presence', which in part is this idea that we stop perceiving the interface between us and the content, and are merely acting and seeing and hearing and being. We have the tracked controllers for being able to give you hand-proxies, but at the moment it is actually game engine tech stack that is preventing use from making those hand-to-thing and hand-to-hand-to-thing interactions feel more solid, feel more reliable, and most importantly, have physics-based interactions at small scales that map to our received experience of the world. Part of this is a perf issue, part of it is a fundamental design/architecture where we need a physics engine tuned for inserting a key into a lock, not a roadster drifting around a turn.

Anywho, I hope this ramble has been at least somewhat interesting/illuminating to those of you who don't do VR dev, and helps to illustrate the degree of technical challenge that we're all embarking on at the moment. Many of these problems are things that disproportionately hit AAA-size developers, as they are behemoths, are slow to change, and are rife with inefficiencies, poor organizational structures, and are frequently crewed with underpaid, overworked, semi-burned out folks who don't want to be given another half dozen challenges on top of what they already deal with. I'm not saying it's hopeless, but I am saying that the big VR games (that really feel like things BUILT for the medium), are probably going to take a while.

# ¿ Sep 22, 2016 12:59

SCheeseman: Apr 23, 2003

Both were rushed to market, they just chose different things to focus on before they pushed it out the door. The Vive's software was pretty rough in places (I think SteamVR wasn't even 1.0 when the Vive was initially released) and the head mounting system and cable feel a bit half baked, but on the other hand it is clearly the more "feature complete" of the two, in a large part thanks to its integration with Steam, the camera and the controllers.

Oculus focused on making the best possible headset, pushing tracked controllers, larger tracked areas and fancy software features to the side because evidently they just weren't ready at the time. The core software experience was more polished, but it's easier to polish what is essentially only a game launcher.

Regardless of Oculus' plans, they should have bundled Touch by default. It was a dumb oversight to not give it a higher priority, tracked controllers are magical and fragmenting the user base like they've done now only serves to confuse end users and frustrate developers.

# ¿ Sep 23, 2016 14:37

SCheeseman: Apr 23, 2003

TomR posted:

Also I want to try the 3d modeling tools in VR.

Someone needs to recreate SketchUp in VR.

# ¿ Sep 23, 2016 15:21

SCheeseman: Apr 23, 2003

There's plenty of other reasons to dislike facebook. It's not like Oculus users are gonna lose out anyway, they'll still be able to play using OpenVR versions of the games.

# ¿ Sep 24, 2016 03:54

SCheeseman: Apr 23, 2003

bloodysabbath posted:

Every developer of software and hardware and every employee of such must be vetted for thier political beliefs and voting history. It�s the only way.

- Posted on my device made in loving China.

Everyone has to pick their battles. If you take your mindset to the extreme, it'd just be pure nihilism.

# ¿ Sep 24, 2016 04:08

SCheeseman: Apr 23, 2003

I'm not sure what Oculus should do. Selling to Facebook was a mistake as the culture there is too consumer hostile for much good to have come from it (apart from the cash injection), at least from the perspective of the end user. I think competition is healthy so it's probably a good thing for them to stick around if only for that?

Basically, what I'm saying is that the whole Palmer thing is more of a tipping point than the primary cause of anger and distrust towards Oculus for many people.

SCheeseman fucked around with this message at 04:28 on Sep 24, 2016

# ¿ Sep 24, 2016 04:26

SCheeseman: Apr 23, 2003

McGiggins posted:

Can't we all just agree that VR is an expensive fad that will pass?

No because that's stupid

# ¿ Sep 25, 2016 04:56

SCheeseman: Apr 23, 2003

I want the Mechwarrior 2 in VR. Specifically Mechwarrior 2, with the graphical style of the DOS version.

Or maybe a VR version of that Battletech arcade machine. I just want a decent Mech game, damnit.

# ¿ Sep 26, 2016 00:21

SCheeseman: Apr 23, 2003

FuzzySlippers posted:

That might actually be somewhat easy or at least as easy as anything is in game dev. I'm not sure what they used at the time but it looks like vertex colored low poly though I think they weren't flat shaded and shared vertices rather than what is currently more fashionable in the flat shaded thing. Looking at the screenshots I'm seeing normals aren't sharp on the edges and colors blend on the faces. I've said it in the game dev thread but avoiding textures always makes 3D work so much easier for me. gently caress UV mapping.

If I wasn't in the middle of a side project about exploring caves and actually had a proper headset yet I might jump on that. Just slavishly copy MW2 with all the names filed off. Like gameplay on OG MW2 was just pick a mech, customize weapons, go on missions about following nav points blowing poo poo up right? Of course you'd end up with a small number of missions compared to old DOS games that would always be absolutely crammed with content.

At the time the engine was a marvel but today you can just shrug through the technicals. Especially if you wanted to be lazy about performance as you'd still be at a bazzilion fps on VR capable hardware no matter how much you hosed it up.

Copying the sound design might be a bit more difficult, though.

Mechwarrior 2 seems like a good candidate for a thorough reverse engineering for an engine re-implementation. It's a fondly remembered game and the engine was used in a bunch of other stuff too, notably Interstate '76.

# ¿ Sep 26, 2016 02:04

SCheeseman: Apr 23, 2003

https://www.youtube.com/watch?v=-X3GD0UnBCk

# ¿ Sep 27, 2016 04:43

SCheeseman: Apr 23, 2003

The more power the better. A 1080 will let you hit higher levels of supersampling too.

# ¿ Sep 28, 2016 03:59

SCheeseman: Apr 23, 2003

When the framerate dips under 90fps, in order to keep from making the user vomit the last rendered image is "reprojected" to match the current position and rotation of the headset. It mostly works, but causes judder and other artefacts.

# ¿ Sep 28, 2016 04:04

SCheeseman: Apr 23, 2003

w00tazn posted:

Should we be making GBS threads all over VR as a whole now? Should we abandon it because one guy turned out to be crazy and then used that crazy to make something magical?
Notch and Steve Jobs both turned out to be shitheads too but noone is boycotting Minecraft/Microsoft or boycotting the iPhone/Apple.

Literally no one was planning on doing that.

People have been wary of Oculus for a while, their business practices aren't particularly consumer friendly and they're arrogantly acting like their the only game in town when they fairly clearly aren't. For most it appears Palmer being an rear end in a top hat was the straw that broke the camels back.

Boycotts of the platform only lightly affect Oculus users as OpenVR is compatible with Oculus hardware (including Touch) by default. No one will lose out on any games, though maybe some of the fancier features like async timewarp and the capacitive buttons on the controllers will be missing. A bummer for some, but not the end of the world and definitely not something that will hurt VR on the whole.

# ¿ Sep 29, 2016 18:51

SCheeseman: Apr 23, 2003

What a surprise, the cheap VR thing is poo poo.

# ¿ Oct 5, 2016 15:06

SCheeseman: Apr 23, 2003

After I learned about the dire motion controllers I didn't hold much hope. The tech they're using simply is not good enough for the job and the only thing that is up to par is the HMD and that seems to have issues with its IMUs.

A few people with high tolerance of simulator sickness will probably be okay with all this, but if it makes everyone else sick this thing is gonna tank. If anyone is looking for VR something I'd just recommend waiting till the prices of the Vive and Rift (with Touch etc) have reduced.

# ¿ Oct 5, 2016 17:31

SCheeseman: Apr 23, 2003

Both the vive and Oculus have very solid tracking. Admittedly the vive can be temperamental around reflective surfaces but 99% of the time things just work. This is necessary for VR to be useful.

The move controllers don't work. They are broken by design and the apparent wobble people are experiencing is really concerning. I'm not gloating, I'm disappointed as while I've understood for a while that it was always going to be an inferior experience I had hoped they would at least get the core head tracking right. What a bummer.

# ¿ Oct 6, 2016 07:19

SCheeseman: Apr 23, 2003

Wouldn't doing that also disallow GPL-licensed code?

# ¿ Oct 10, 2016 01:47

SCheeseman: Apr 23, 2003

It's not official but I've played quake in VR and it made me feel like poo poo afterwards.

# ¿ Oct 12, 2016 20:02

SCheeseman: Apr 23, 2003

Yes everyone should pay attention to Alex St John, he has very interesting things to say about the industry such as this: http://www.alexstjohn.com/WP/download/Recruiting%20Giants.pdf

Edit: oh wait he's an irredeemable rear end in a top hat

SCheeseman fucked around with this message at 08:11 on Oct 17, 2016

# ¿ Oct 17, 2016 08:09

SCheeseman: Apr 23, 2003

gonna buy I dildo v1

# ¿ Oct 18, 2016 14:38

SCheeseman: Apr 23, 2003

I can't wait to enter the Mega Space Hole in my Dildo v1

# ¿ Oct 18, 2016 14:40

SCheeseman: Apr 23, 2003

I think of room scale virtual reality as a sort of stepping stone towards augmented reality. Positionally tracked AR is really just VR with an alpha channel of sorts, though thanks to that relatively minor difference AR has quite a bit more potential.

Even basic stuff like tutorials, or checklists, instead of reading instructions off a book, you could simply know what to do from nav markers in (what your mind perceives to be) your actual physical space. Object recognition, image enhancement like brightness, contrast, infrared, zoom; think predator vision (except not blurry as gently caress, hopefully). Imagine being able to accurately judge the temperature of something just by gazing at it, being able to summon arbitrarily sized flat screen displays, any aspect ratio you want with quality levels exceeding any LCD today. Once AR glasses get small enough it's inevitable that swaths of electronics will become obsolete, particularly anything with a screen attached to it.

All of that has to start somewhere: with a head mounted display that despite all it's flaws manages to do a good enough job of replicating human vision to the extent that it's believable and allows the basic functionality to be able to interact with the virtual world in an intuitive way. It's a necessary first step that needed to be taken in order to be able to figure out the basics on how to exploit the new medium.

# ¿ Oct 23, 2016 18:41

SCheeseman: Apr 23, 2003

Here's a good example of possible real world use in the future: http://store.steampowered.com/app/357670/

This a fun, dumb little game. It gives you a boxes of flat pack furniture and some basic instructions and lets you go to town but consider what would happen if they actually took advantage of what VR could do to guide you instead of mimicking real life.

What would happen is that the game would be too easy.

# ¿ Oct 23, 2016 18:55

SCheeseman: Apr 23, 2003

So Microsoft are starting to push Windows Holographic.

quote:

Microsoft�s partners will ship a new line of virtual reality headsets to take advantage of Windows 10�s VR and holographic capabilities. At today�s event, Microsoft said that the headsets will start at $299 and will include inside-out tracking sensors, obviating the need for external cameras or laser systems like those on the current Oculus Rift or HTC Vive. HP, Dell, Lenovo, Asus, and Acer are all listed as partners.

While relatively little is known about the headsets, the six-degree-of-freedom tracking system sounds similar to Oculus� Santa Cruz prototype, which also includes inside-out tracking. They are distinct from Microsoft HoloLens, an augmented or �mixed� reality headset that projects virtual objects into the real world. While the headsets don�t need external trackers, the one we saw on stage today still looks to be wired, not totally self-contained like Santa Cruz.

I think they're likely using the tracking system Hololens uses. From what I hear it's pretty good, though they didn't demonstrate nor announce any 6DOF controllers. While hand tracking is cool I feel that you need some kind of tactile feedback and despite people claiming that you don't need to know the position of your hands all the time, losing tracking of them when you're not looking at them breaks a lot of cool game gestures and mechanics. Also no word on whether these headsets will support OpenVR though I think it's likely (conversely the Vive will likely support Windows Holographic). Oculus, notably, was not mentioned.

# ¿ Oct 26, 2016 18:13

SCheeseman: Apr 23, 2003

Considering that the model demonstrated was an all-in-one unit, a big pricing advantage would be in component cost. The Rift, Vive and PSVR are kits with the cameras/lighthouses, controllers and cables all adding considerably to the weight and cost of the unit. Selling just a headset with maybe a single tether is going to be cheaper, even with the addition of 3 extra cameras on the HMD.

It's a smart move. I'm still interested what they're doing with input, but it's possible to do a hell of a lot more with the technology they're promising than a stock Rift. A good deal of Vive games could conceivably work with purely hand tracking inputs too and if they follow the trend of adding ASW like the others, it may well be possible to put together a capable VR PC at less than the cost of PS4+PSVR.

# ¿ Oct 26, 2016 20:37

SCheeseman: Apr 23, 2003

Microsoft hardware has been generally quite good, the 360 RROD debacle notwithstanding. People love their Surface Pros and their peripherals are excellent.

# ¿ Oct 26, 2016 22:37

SCheeseman: Apr 23, 2003

bloodysabbath posted:

The Xbox One would like a word.

The xbone is fine it's just that the PS4 is better.

Honestly, in my case I think that I'd only have use for the former due to the 4K BluRay capabilities. Maybe. When I get a 4K something.

# ¿ Oct 27, 2016 04:40

SCheeseman: Apr 23, 2003

I gotta wonder what the chances are that those headsets get OpenVR support too. Seems like a no-brainer but I gotta wonder if licensing tracking tech from Microsoft comes with a few restrictions on implementation.

# ¿ Oct 27, 2016 07:03

Adbot: ADBOT LOVES YOU

# ¿ May 13, 2024 16:25

SCheeseman: Apr 23, 2003

Vince MechMahon posted:

So I picked up a PSVR today. Is there currently any way to do 3D movies in it?

Weirdly, no. It's capable but they haven't released anything that can do so.

# ¿ Oct 27, 2016 07:21

The Something Awful Forums > Discussion > Video Games > VR Megathread: Valve's VR Has Entered The Game

«‹›16 »