Biggest Issues Affecting VR

Just for my own understanding, what would you consider being the biggest problem with VR today? Why is it that developers and consumers need very expensive machines in order to run their software? What would developers consider the biggest performance issue they must overcome?

Im trying to get into the VR development space and would like to be thinking about these things as I develop my software. Thanks in advance.

 

I'm developing at the moment and looking to exploit the mobile market. Theres a LOT of small apps but nothing with a true story or structure to them.

Ideally my aim will be to provide some form of cross play later in development, but as it stands, most people have mobile phones, and they are very capable now with Cardboard / Daydream! A lot of the problem comes from the fact you are rendering the same scene, twice. Hence the need for massive hardware!

 

1. price
2. (consequently) market size
3. (consequently) lack of games (just meme games and 5 minute demos)

 

Terrible graphics quality compared to the mobile, console and PC markets. VR has jumped back 20 years in graphics quality. We need to figure out how to optimize throughout the full stack for VR. I’ve done dozens of demos, both individual and at trade shows, and graphics quality is always a let down.

I think multiplayer/social aspects haven’t been worked out really well yet either. Right now it’s too isolated.

Setup is intensive too, too many cables.

There hasn’t been a blockbuster game. There have been niche hits, but nothing like a Mario.

 

Mass consumer adoption just didn't happen for VR devices. It's a shrinking niche. By the time standalone daydream devices are released, who's gonna pay $1000+ for something like that? Very few people, a lot less than Rift or Vive which already has low consumer numbers and not growing.

 

I was legit interested in oculus... when it cost under 500 and worked on my under 300 GPU. (development kit 1, I think?)

But even theh it was just because I'm into combat flight simulators. I even build a cockpit ~5 years ago and mounted a TV on it.

Aaaah those calm days of beer and Sturmovik...

 

Well.. Oculus Rift + Touch controllers are only ~$399/450€ at the moment as there is a summer sale going on.

 

Actually the Rift Market is growing. As a result of the sale all Rift units are sold out and on backorder again. No numbers were given on how many units were sold but that is a promising development in market growth. After the sale the Rift is dropping to 500$ with Touch controls. That also lowers fragmentation between non gamers with or without motion controls since we now know all recent purchases and future purchases only come with motion controls. The Vive with 300$ dollar price above the Rift is sure to trigger a drop it as well if they want to stay competitive. I would say the biggest problem facing VR is game journalism giving negative press.

What we can do as developers in my opinion is make our game's meatier in length as that is the number one complaint I hear back from people waiting to jump into VR.

 

It's certainly growing, but will still be niche for some time.

https://uploadvr.com/psvr-vive-superdata-sales/

The rift/touch pricedrop will help but they were getting trounced by the Vive and had to do something. I wouldn't be surprised to see either htc drop the vive's price or other cheaper Steam VR headsets appear. That being said positional tracking VR cheapest option is PSVR at ~$750 (playstation + PSVR). The PC options cheapest option is ~$1000 (rift + high end PC) those are way too expensive to not be a niche market like @Selzier said. In a few year when the total pricetag gets to around $400 then it will have a shot.

Market data is pretty clear that cardboard has the most but for games it's subpar. Gear VR has 5 million + users and the others are all around 1-2 million. Full featured games that offer many hours of gameplay are still hard to find so until that changes I really wonder how many of those sales are just collecting dust.

From a developer perspective Gear VR seems to be the most healthy at the moment. The number of reviews on titles indicates there are quite a few purchases, much more than rift/vive. Daydream is picking up the pace, but the play store lists total download ranges and even the most popular games have a low number of downloads.

 

And my GPU is now 7 years behind recommended specs.

Got any examples? I'm kinda curious what they write these days.

 

I'm working on a story driven, procedurally generated game at the moment to try and explore the cardboard market, which seems to be the best option at the moment. Problem is it is an absolute pain to provide any form of fidelity, at a decent framerate. As soon as you use the GVR event system CPU time skyrockets.

 

Is mobile good for anything other than 360 video?

I wish there were stats on the revenues of the occulus store, steam vr, google vr, etc. I might be totally wrong, but I would assume that mobile vr doesn't consume a lot of content. The quality just isn't there. It's hard enough to create quality content on Vive and Oculus.

 

Daydream is struggling a bit, but they got started later than others so we'll see. Gear VR seems to have the most active users, more than both steam vr and the rift. If you are talking strictly about revenue I would guess that gear vr is roughly equal to both PC options (at least double the users but the games are typically half the price of their PC counterparts). I'm basing this on the number of reviews on titles that are available on both platforms. Actual sales numbers as far as I know aren't available but you can get a rough idea from the number of reviews per title.

As far as the mobile landscape goes from a functionality perspective, your statement seems to line up with the cardboard experience. Gear VR and Daydream with the motion controllers, while they aren't positional tracking by any stretch, do make some decent games (eclipse: edge of light, virtual virtual reality, smash hit, minecraft, dead and buried, etc.). The daydream developers thus far seem dissapointed in sales (but they are exclusives so perhaps google compensated them as well) but gear vr seems alive and well.

 

Ive seen games that coupled with a bluetooth gamepad do work well! Galaxy VR being the one that springs to mind. The one thing I've noticed is that they severely lack depth!

 

I built GalaxyVR using ShiVa3D engine, I'm just one guy. Yes many mobile VR games lack depth, reason being... true artistic/interesting depth is very expensive to create requiring many man hours of work.

When developing games, as scope grows: the work increases exponentially, things have dependencies on other things. You want to change a core mechanic and suddenly you have 100 scenes that need tweaks. For this reason the saying "the last 20% of development takes 80% of time" is true if you want a commercial grade product, certainly when no procedural techniques are employed. This can be solved with brute force by throwing ridiculous man hours at the problem, as the AAA studios do. But VR userbase is too small for that expenditure, no chance of profit.

I think as indies, to make "deeper" games we need to use cheats like procedural generation as you are thinking about. Had it not been for procedural gen I couldn't have built and play balanced/tested 100 3D levels by hand.

Or we can make online multiplayer experiences with revisit appeal created by real humans playing together online, I'm trying this next for Daydream/GearVR:

Spoiler: MechZ

I do not expect my income (even if successful) to be any where near that which I earn in employment, not until the userbase gets bigger. I'm coding for less $ than I could earn flipping burgers.

It's kind of a gamble on the future. Will/Can VR growth go exponential to mainstream?.... If it burns it will, if it fizzles it wont. Form factor will help adoption, when rappers begin wearing XR raybans we'll know it's mainstream.

 

Well said

 

I've always wondered why AAA studios don't port some older desktop titles to VR. Surely some titles would be a fairly simple port. Some great "deep" games made in late 90s that 2017 phones could chew through without even overheating. Currently folk often have to put up with low poly untextured models in VR, so professionally textured low poly models are an improvement.

 

Nausea. Period. There's no bigger issue than this for the VR market. The biggest barrier for those who first try the headset with any game or VR experience is: most people tend to feel sick, no matter how hard a dev may work on avoiding the issue. The problem, as I was mentioning in another thread, is the hardware itself. The biggest reason people get so sickly in VR, I think, is because the camera can jitter or move too fast, faster than real life, (since it can move at the slightest move of the head). in contrast, our eyes actually stabilize the image by moving the eyeballs separately, focusing on the target and keeping the image generally stable, even if the head is moving.

While VR headsets track only the head, and not where the Player's eyes are focusing, the great majority of people are going to keep feeling funny. Else, we could use the head for position and the place where the eyes are focusing to control the target of the camera (like an anchor point during head movement)... stabilizing the image, like in real life. That might feel more natural, and less nausea inducing. I do think people can buy upgrades for eye tracking. Not sure if it's being used at all, though. I did just post about this to Adam Myhill, so maybe he'll be able to test this (he's a Camera wiz too).

 

I'm curious: what VR platform/s have you tried that manifested jittery tracking?

RE EyeTracking: You are misunderstanding things slightly.
When a user moves their eyes the scene camera should not be moved.
When a user moves their head the scene camera should be moved (ideally with 0 latency).
The users eyes are free to scan around the scene, just as in real life.
If you move the camera depending on eye track it would be an awful experience.

Benefits of eye tracking in VR are:

• Foveated rendering.
• Possibility of a gaze/blink interface.
• Analysis of users attention.

I can't think of any more.

 

That's not really what I meant, I just meant that in real life moving your head doesn't change your focus point much. It's hard to explain, or maybe it's just me. I guess I do not have enough experience to give an opinion, since I mostly program the gameplay in my current project. Still nausea is one of the biggest issues right now... keeps a lot of people away from VR.

 

There's no problems with preventing VR progress at the moment. Where we are at with VR could be considered similar to the 8 bit era for traditional gaming in relation to today - you'll need 30 years before it is fully realised without the headset nonsense I imagine.

So this is the infancy, the start. It's a fun time if memory serves - full of innovation and research. Enjoy it but accept it will be limited by price, software and hardware for a long time to come. That's part of the journey.

The visuals will likely be improving a lot within 2-3 years, multi res shading, dramatically increased bandwith and so on. But it will always - without question - lag behind single renders, obviously. VR has to run at really high framerates and render a lot more. There's no fix for that other than good design.

An immediate problem is VR movement, this is pretty hard to do without your testers throwing up and the teleportation style sucks - it breaks immersion a fair bit and you resort to blinks or portals.

 

OK I see what you mean now. Yes VR at the moment has the whole image presented at fixed focus. Indeed eyetracking can analyse users Z plane attention and focus could be adjusted accordingly. That would probably reduce the %ge of folk getting VR sickness. In fact I think there is a research system out now that uses DLP projection chips and that can pull the entire scene focus depending on gaze. Saw it on "Tested", cant remember system name.

 

I think this sums it up perfectly. I also don't see any reason for VR adoption to stall.

Progress/adoption will have exponential component to it, just as home computers have.

Were I to time travel back to 1981 and tell my 9 year old self (who had a ZX81 computer) the specifications of my 2017 laptop.... Well it would FAR exceed any expectations or predictions by orders of magnitude.

Human brains are wired for linear predictions, not exponential, sure consciously we may understand them, but they will still surprise us every time.

Before mainstream adoption it will be called Extended Reality XR: viewing devices capable of AR or VR using some layer that can have opacity toggled. A premium pair of XR shades in 2045 will cost <0.001 Bitcoin.

 

I agree to all the above, but you got to admit when people first played Nintendo or something like that, they didn't usually get sick from it. Some people are a bit wary of continuing to use a VR headset when they notice it makes them feel nauseous (and I'm talking about roomscale and movement). This is specially true of people who don't own a headset and just tried it at a friend's or at a store, whereas a headset owner will have "buyers guilt" pushing him/her to continue using it, since they paid a lot for it...

Anyway, I guess we can just wait and see where VR will go.

 

Once a pastime becomes popular many apes follow regardless of physical consequence, e.g. drinking Ethyl Alcohol. But you're quite right: Ideally there should be no incidence of sickness.

My own experience is limited to seated VR (rotational head tracking only). With some thousands of users I've had no complaints about motions sickness, and I do rotate/translate the camera around (in a cockpit). Of course my sample set is also prone to selection bias; my prospective new users tending to be already familiar with phone VR.

I've heard AR causes virtually 0 sickness, makes sense as user is in real world. So XR will probably be a good way for any user to enter VR at their own pace. I can imagine a future XR demonstrator app whereby user sets a dial 0 ... 1:

0 = Full real world.
>0 ... <1 = some level of AR
1 = Full VR

Everyone would be fine at 0, it's just a pair of clear glasses.

Related: Some have complained that google daydream "view" headset has too narrow a FOV. I suppose It does but because of that it also has a lower incidence of sim sickness. That maybe another way to reduce sim sickness: user defined FOV, user slowly opening up the FOV over some time as they gain their "VR legs". Application level implementation might use simple scale able mask meshes (quads with holes) in front of each stereo cam or something.