Hacker News Reader

Comment by rwz | Hacker News Reader

I've never been interested in VR headsets and am not very educated about it.

Can someone explain why you'd want to actually run things on it instead of having a much lighter screens-and-headphones-only unit that plugs it into a more powerful machine that renders things for it? What's the benefit of having to strap all this extra hot hardware and batteries to your face.

Is it just about it being wireless? Surely there are wireless protocols that can stream high-bandwidth videostreams from your PC to VR goggles.

Not trying to be snarky or facetious, but genuinely curious.

View on HN

I'm surprised no one really has answered the implied question. Wireless protocols are hard for VR not because of bandwidth, but because of latency. Head tilt motions have to be processed extremely quickly to not be nauseating. Even with wired systems, the headset often ends up doing all kinds of motion prediction to process motion and shift the image before the game has time to process it, because engines aren't going to round trip input to video output at 120fps/8ms (double buffering doubles your latency!).

Of course, that multiplies with bandwidth requirements, which are also quite extreme at 4k 120fps.

To be clear, you don't have to run things on it, you can stream wirelessly from your PC.

An advantage of a standalone unit is you can take it with you, or for people who don't already have a gaming computer.

You're also asking why not have it plug to a more powerful machine. The answer for that is simple: wires get snagged and pulled on, in practice it's a poor experience (remember, you cannot see the wires when you are wearing the headset).

> To be clear, you don't have to run things on it, you can stream wirelessly from your PC.

But the hardware is adding weight and bulk to something that you really, desperately want to be as small and light and possible.

Well, there are already devices like that (Big Screen Beyond), Valve has a wired headset (Valve Index) - so it makes perfect sense to do something more modern and future proof.

The Index is almost a decade old and has pretty crappy screens compared to what they could use today.

The real question is why did they release this as a standalone unit for $1049 and then claim they couldn't sell it for less due to the RAM shortage - when that implies they could have released a dumb unit that depends on the Steam Deck or Steam Machine for far less?

If it can do innside out tracking its way more data efficent to have compute onboard. And if you already have a decent amount of compute you can scale it up just a little more to have. a fully standalone unit.

I'm not sure going from less than $50 of RAM and a $50 processor for foveated eye tracking to 16 GiB of RAM ($300) and a Snapdragon 8 ($200) is a "slight bump".

Wait I don't realise ram prices where this bad. Bought my kit for about a tenth of that a while ago. Ouch.

Yeah, can't wait for the AI hype bubble to finally end. Hey, we might even get some nice fast repurposed HBM from the scrappers reusing stuff from the bankruptcy auctions. :)

Yeah! Just like we all made out when the cryptocurrency bubble burst too!

Yeah! Got some nice and very cheep GPUs back then. :)

Foveated eye tracking, but also SLAM for the controllers. Also, to be fair, the onboard processing costs less than a separate VR capable PC.

The onboard processing costs a whole lot more than a PC you already have.

compute is fine. its necessary battery is not. compute is almost weightless, battery crushes heads.

Which is why it is in the back part, balancing out the frot part.

Balance is only one thing, total weight still matters very much.

Even with some compute on board you could save most of the additional weight by making it tethered.

You're asking why they couldn't completely change the product from a full-fledged VR machine into a headset in roughly a year, especially when they already sell a regular VR headset for $1,000?

They do not, in fact sell such a headset, in that it is permanently out of stock. You can risk buying it used for $1800…

It was released in 2019 and by comparison, sucks. A new index with new displays of the frame but without the things that make the frame expensive (ram) would be welcome.

It still shows the headset in stock in the US. The bundles, controllers, and lighthouses seem to be out of stock though, which is likely because they're a much more popular product (the controllers/lighthouses break quite often, and they're often used with other headsets as well).

Valve Index Headset Only: "Requires Base Stations and PC." - In stock.

Valve Index Base Station: "Out of Stock"

You cannot buy a functioning Valve Index. And you wouldn't want one, because its 7 years out of date.

They've doing inside-out SLAM to map the environment and tracking your eyes for foveated rendering at the same time. That's fairly heavy - I think the compute required would necessitate a much higher speed connection with a desktop and a fair bit of CPU time, so it's probably cheaper and easier to put the compute right in the headset instead of scaling up the connection.

Also, they probably landed on this rough design a few years ago - before the AI hyperscalers bought up all the RAM - so the economics changed a lot under their feet.

Maybe I'm misunderstanding - but I thought there's already "dumb" headsets doing foveated eye tracking, like the Pimax headsets. If that's the case, then you don't need $300 of ram and a $200 CPU to perform this locally.

You're misunderstanding.

The Pimax headsets with foveated eye tracking start at $1599.

You don't need $300 of ram and a $200 CPU to perform eye tracking, otherwise the Steam Frame would be able to do nothing else

If you read the chain of comments again you will notice that eye tracking is only one of several things that it is doing.

why do you need 16GB of RAM for eye tracking and SLAM?

I understand that thinking is difficult now that everybody offloads their thought to LLMs, but the math is very simple.

The cheapest comparable device with the same (or better) level of eye tracking is $1500 (i.e., 50% more than the Steam Frame)...and that was the price before the tariff war and Iran conflict jacked up prices.

While the ram may currently be $300, the CPU is not $200 unless you're including the custom processing units Valve is using to handle the eye tracking, which are (a) industry leading, and (b) made to spec in relatively small quantities and therefore more expensive (and given the limited chip capacity also affecting chip manufacturing, also significantly more expensive than they would normally be).

If foveated rendering was cheap, it would already be standard in every VR device.

If Valve took the Steam Frame and removed: The SoC, memory, storage, battery and power system, wireless function and added a display cable, would it:

A. become ~$300 cheaper, close to the PlayStation VR2 which also has eye tracking

B. magically go up in price to $1500 like a Pimax headset?

If answering is difficult try asking an LLM for help. You're their target audience.

I like wired for the better graphics and unlimited playtime. In practice even a single cable gets wound up, stuck on things, and generally feels in the way. It's a difficult trade off to accept, even with a long and light weight cable.

Being able to play without a bunch of setup in your room and without worrying about getting tangled in cables seems great. I have a Valve Index—which is a pretty remarkable device, that I definitely enjoyed—but the hassle of cable management/etc adds some real hassle. And I say this as somebody who actually set up a pulley system to make it work better in my home office :P

The wires really get in the way, even in a seated context. I often use a Rift S for rally games, and the wire will get caught in the backrest, the wheels, my legs. VR is immersive, so you instantly forget and it's a thick and fragile wire so there's a degree of stress.

Not to mention the trip hazard or ripping it out of the PC while playing room-scale games.

I concur, given hardware constraints that a far more slimmed down unit would have been appealing, and more future-proof, but I believe there's still a requirement for onboard compute for inside-out tracking and handling disconnection events, at some point making it self-sufficient (at the time) was seen as not much more expense.

Unless you are standing still (like sitting or playing Beat Saber), that cable really is annoying. You can't spin around for instance.

As for wireless communication. I also thought it would be easy, but the lack of options despite the demand in the Oculus Rift / HTC Vive days proved otherwise. There have been some attempts, some worked, none of them great, and they usually involved hot hardware and batteries.

Steam Frame is apparently the best at being a wireless, PC-driven headset.

As for standalone functionality, my guess is that a headset hardware powerful enough to deliver a good wireless PC VR experience is powerful enough to work as a standalone headset for less demanding applications.

I had a TP Link wireless adapter (I think that was the name) for my HTC Vive. I was kind of skeptical going in, but it worked shockingly well--no detectable added latency. The only thing I lost was the ability to use the Vive's camera, which I think was just a software issue of some kind.

There are two use-cases. Already sitting at a PC and you just pop your headset on - your idea makes sense. But traditional VR means standing up in a dedicated space (2m x 2m) and demands uninterrupted time. That's a high friction activity, and adding setting up a PC and Oculus Link and Steam VR before you start playing makes that friction even higher to the point where it's a niche hobby. Most casual players are doing stuff like VRChat and Beat Saber which can run on low powered hardware, so having a basic game-playing ability is a good idea.

At least right now, the "mobile" version of VRChat is not that impressive (well, its impressive what can even run when you cut it down this much).

All the really impressive Worlds and avatars are still PC only.

Ideally, there would be some more seamless performance scaling than the current "do 2 versions, one of them ugly but fast, if even possible". :P

Maybe Steam Frame will help this get going by being in a better place to run more advanced stuff, being more powerful than Qeust 3, which was basically the "mobile" VRChat benchmark so far.

1. Unless all you play is a sit down driving or flying simulator, the wires will get in your way. You can mitigate it with ceiling and wall pulleys, but it's still a really terrible experience that kills immersion.

2. Fitness. If you're moving and turning a lot, you're just going to hate wires. Your arms will slam into them and you'll almost trip on them and pull your computer down.

Wireless to a desktop isn't perfect, but it's doable. Stand alone is ideal if you don't the visuals being turned down a bit.

It's a lot easier for developers to target a known platform.

When you’re in VRChat you can be your fursona

It is designed to stream wirelessly from a powerful computer too

I’m shocked that this is performant enough to work in practice. Isn’t the timing window for VR responsiveness extremely tight?

It works, but there is usually a noticeable difference (at least with Quest).

I have a dedicated Wifi 6 (not 6e, so just miss/using the 5 GHz band) AP just for VR, directly connected to my PC and it works well Virtual Desktop & Quest 3.

Have done quite a lot of quite active stuff (exploration, driving, flying, fighting games) and it really is seamless and I don't notice any lag as long as the game itself manages good FPS.

If you really concentrate on the graphics, you can see some compression artifacts in places, but it is unnoticeable when you are actually doing VR stuff, not just looking for it. :)

For the Frame, this should all be much better - easier setup, much better available bandwidth (dongle has Wifi 7), efficient foveated streaming, etc. :) I would exect better image quality (more bandwidth, more efficient streaming), better FPS (more efficient rendering) and better image quality (can reduce compression).

Good to know!

They are using foveated streaming where they track where your eye is looking and only stream that part at full resolution.

They also have a dedicated wifi stick that directly connects to the frame in addition to using your normal wifi network.

I was more concerned with the latency than the bandwidth.

Bandwidth is very important for low latency with wireless VR. To have a low latency you need to give the VR headset a video stream of at least 90 fps. That means every 11ms the headset needs a full screen frame of video. Over regular Wi-Fi 5 from my phone I can ping my desktop in 4ms so is the latency problem solved? The problem is there is not enough consistent bandwidth over my consumer Wi-Fi router to maintain a high quality experience in VR.

Enthusiasts solve this problem with a dedicated VR only Wi-Fi router connected to their PC but now it will be much easier with the included 6Ghz Wi-Fi dongle in the box

In the Tested review, they got 5ms latency with the dedicated wifi stick! I believe that was with 250mbit streaming. If they set it to max streaming bandwidth, it was 50ms or so.

I suspect the answer is actually very simple, when you compare this device to one that is made deliberately dumber there are far too few actual differences to make it make sense.

Reply on news.ycombinator.com