Meta Was RIGHT To Skip Eye Tracking on Quest 3 | Episode 16
ฝัง
- เผยแพร่เมื่อ 23 มิ.ย. 2024
- On this Live Open Daily, I double down on my take that Quest 3 should not have had eye tracking, in spite of compelling reasons from Apple and Sony saying otherwise AND a new third party solution on the way.
Subscribe at / @liveopenmike
Join this channel to get access to perks:
/ @liveopenmike
Live Open Mike is a virtual reality gaming channel with list videos, VR game reviews, VR gameplay footage, and VR gaming news. For full VR livestreams, check out Live Open Mike on Twitch at / liveopenmike
Follow on Twitch at / liveopenmike
Follow on Twitter at / liveopenmike
Join the Discord at t.co/0xkWhpsQiA
VR Gaming Merch! liveopenmike.creator-spring.com
#metaquest3 #virtualreality #vr #metaquest2 - เกม
good video. The background music is too loud.
@@viperjay1 Thx. Still working on getting the levels right.
Yeah, I was just about to post the same thing. Soooo annoying, I can hardly concentrate on what is being said 😢
@@alanharrington Poll coming this week. Make sure to vote.
Same, it was too load. Good content though!
@@ruudboek Yeah, taking all the feedback on board. This is the first episode of the pod to get music complaints, but I screwed up on the sound leveling, so... yeah poll on the channel tomorrow about whether or not I keep music or not.
Hey look who I found in my eye tracking research :D
Algorithm ftw! My eye tracking video from last year pissed a lot of people off, so I doubled down. lol
My perfect headset would be eyetracking, pancake lenses, bright Oled screens of psvr2, size of quest 3. Hopefully most these features are in quest 4
I don’t think Meta goes OLED until production yields get better. Sony and Apple couldn’t produce a million units in a single production run for their headset and low production yield equals higher pricing.
All that and you either have 1 hours of battery life or a very heavy headset (or an external battery)
It doesnt have to OLED - ill take QLED of Quest PRO and like 4x the dimming zones, that be a good stepping up and WAY better then standard LCD we got now
@@mastorasmars2326 That's more feasible, yeah.
The other cost saving benefits of DFR benefit the end user in the long run. DFR is already known to make a decent improvement on performance which reduces the compute on the device or host PC. From the user's perspective as far as PCVR is concerned, this means not needing to buy as beefed up of a GPU and allowing lower tier GPUs to perform better in VR (which makes VR more accessible without needing to spend an obscene amount of money on a gaming PC - it lowers the cost of entry) and from the Quest 3's standalone perspective - this means less heat and more resources to play with. The reason the Q3 struggles to stay charged during use is because Meta has to dial back the charge wattage as the device is running hot. Increasing its overhead with 10-15% extra overhead is hardly minimal when you have people overclocking GPUs and CPUs for much smaller gains and would take some of the strain off the hardware and reduce that heat allowing it to charge more efficiently whilst in use.
If hardware does not have the tech then that is not going to persuade software devs to implement the feature more. When it comes to the Quest 3, I am pretty sure the XR2Gen2 SOC is capable of running 10 concurrent camera streams (Quest 3 has 4 on the front for the passthrough and 2 for controller tracking on the lower edges of the headset as well as the MR camera). This means that the Q3 would have had the bandwidth left over for one camera per eye and a camera for the face tracking as well even if the face tracking itself is purely for social VR use, it makes eyetracking more attractive to the user and more likely to upgrade if they are getting both rather than just eye tracking. Pico 4 Pro for example has Face and Eye tracking though it is only available outside of China through sites like Aliexpress, is much more attractively priced and accessible than the Quest Pro.
The hardware needed for eyetracking is not expensive. Have you looked in to the open source project EyeTrackVR? It uses a couple of ESP32 boards, some inexpensive camera modules and some low power IR LED emitters as a DIY solution to bring eyetracking to any PCVR capable headset. The fact that the Q Pro already had the infrastructure and software in place for both eye and face tracking on both PC and standalone, it would not have been expensive for Meta to build this in to the Q3. The reason the Q Pro failed is nothing to do with the Eyetracking and everything to do with the way Meta tried to prematurely shoe horn VR and MR in to the business space and charge a premium for it to customers that didn't want it instead of concentrating on the target market of VR which is still gamers. The removal of ET and FT from the Q3 was purely a shortsighted cost saving exercise of pennies that if Meta has included these features, it would have persuaded a lot more people to be willing to upgrade to the Q3 as it adds value from not only a social perspective but also the ability for devs to implement features such as DFR in the future if they see there is more mass user adoption for a feature that would make their game or app better than the competition. It would have given devs the power to add further value to their products too by implementing DFR or social ET and FT capability even if this was in the future once any teething issues with SDKs or dev requirements were ironed out more.
Additionally, the SOC in the Quest Pro was already outdated when it released and with a resolution barely better than the Q2. Meta was charging a steep markup for dated tech and pushing it to an industry that isn't ready for it yet.
All the headsets you mentioned in the video are higher end market devices that most just don't have the means to access and come with their own issues.
That is my take at least.
1) Meta DOES NOT CARE about PCVR. I laugh whenever someone brings up the benefit for PCVR players because Meta has made it clear.
2) Meta’s highest priority is price as they know they’re in the value consumer market, not the enthusiast market. I believe every Quest headset will have a major sacrifice for the sake of pricing.
3) Until more developers are actively using DFR and have upgrading their engines to do so, it doesn’t make sense to add a low usage feature. I think this happens in a cycle or two, but for Quest 3, the timing wasn’t right.
One important thing to consider is that eye tracking doenst just add cost but also weight (sensors), performance and battery power to run all these sensors. We saw with the quest pro that while using the sensors, batterylife and performance suffered.
This doesnt matter as much for pcvr, as the performance needed for eye tracking is just a fraction of the power budget. This is different on stand alone.
Simplified example:
PC: 20 Tflops performance
Quest 3: 2 Tflos performance
Eyetracking needs 0,5 Tflops to get processed and gives a 30% performance boost.
On pc: (20 Tflops - 0,5 Tflops) * 1,3 = 25 Tflops Performance
On Quest 3: (2 Tflop - 0,5 Tflops) * 1,3 = 1,95 Tflops.
Adding to this, the quest 3s soc is already underclocked to keep battery life at a certain rate. Using power on eye tracking sensors as well would probl require the soc to be even clocked lower.
In the end you might just get more performance with a higher clocked SOC rather than adding eye tracking for DFR.
@@liveopenmike They might not care about developing games for PCVR themselves but if they truly didnt care about PCVR they would not have implemented Link or Airlink for the Quest 3. So why did they? Because they know that their users most likely use the their hardware as a PCVR headset either through Link, Airlink or Virtual Desktop (which earns them profit as it is a paid software). Additionally I gave examples of why ET benefits both PCVR and standalone.
I think you missed most of what I was saying.
Additionally.. for your 3rd point it makes no sense. why would a dev bother to code for a feature on a device that does not exist. It would be like coding an RGB software control app for a device that has no RGB lighting to control. The capability needs to be there and for it to be common place enough in the industry that it makes it worth the while of the devs to implement it. And because of the attractive pricing of Quest devices (comparatively to enthusiast level headsets) they are best placed to help drive this adoption within the industry.
@@tenshik3769 Air Link & Quest Link exist to add another use case for the headset, but beyond that program (that Amanda Watson had to push to be allowed to develop when she was still at Meta), Meta's shown little care for the medium as a whole and certainly aren't actively pursuing games for it.
I think people need to let go of this idea that Meta is going to innovate push the boundaries of XR tech and always make the highest end headsets. That hasn't been the case since they bought the company. All of their headsets have had major compromises for the sake of price or form factor or whatever.
I hear what you're saying, but Meta's more interested in making Hondas that everyone will drive, not Teslas that ony the top 10% of the consumer base can afford and covets. That's not their business model.
So they'll add eye tracking when it's more affordable and can scale properly, not before. I'm all about managing expectations here. It sucks because they are the market leader, but that doesn't mean they're going to always have top end gear. They'll always make the jack-of-all trades just good enough to work for everyone headsets.
The high end will always belong to the Varjos and Pimaxs of the world. It's sad, but that's the reality.
@@coolertuep I cant comment on the examples you gave regarding performance but I can say that the weight of two small IR cameras and some IR LED emitters is hardly a weight consideration. The Q3 is already 207g lighter than the Quest Pro and only a couple grams heavier than the Quest 2 but because of how much slimmer the device is, the comfort is miles better than the Quest 2 so this isnt really a perceptible difference. Additionally, the XR2Gen2 SOC is already there and running so no need for additional ESP32's like with a DIY implementation, you would only need the cameras and the emitters. the camera modules for example used for EyeTrackVR only use 140mW each. So if we assume two for eyes and then one for the face and then a couple IR emitters.. We are talking a small additional amount of overhead that could have been fixed entirely if they had not placed the battery in the front of the headset which would allow for better thermals. Heck, even if they made it so the device can run from power directly to relieve the reliance on the battery when plugged in, this would have helped.
If the quest 3 had eye tracking good enough for foveated rendering, I would have paid an extra $400 for it.
At that point, it would have been a feature complete headset that I wouldn’t feel the need to upgrade soon rather than a stopgap while waiting for the actual headset down the line.
I believe you’re in the minority though. Even $500 was too much for a lot of Quest 2 owners, hence why Quest 3S is coming.
@@liveopenmike I believe you. I’m just tired of buying brand new headsets while already thinking of the next headset I’m going to get later.
It reminds me of when I was a child and I had my flip phone, a small camera, an MP3 player, and a GameBoy, all stuffed in my pocket. I could just see the future and it made me upset every time I had to buy something that was not that. When the iPhone came out, I was glad to pay what was at the time a lot of money for a phone.
i always feel a ring around my pupils afterwards or right after i'm done
Try dropping your brightness a hair?
Eye tracking is important.
I agree. Still didn’t make sense for Quest 3.
Lets get Quest 4 with face tracking - QLED at 2.5-3k per eye with 2k dimming zones per eye - latest snapdragon - double resolution passthrough - and if cant manage that with Quest 4 then make it a Quest 4 Pro for ones who wanna spend more for better tech then entry level stuff
I think we'll eventually get another Pro level Quest headset, but that first one bombed. lol. Meta's trying to straddle this line between making a high end headset, but still making it somewhat affordable and that catch 22 means they'll still make compromises on it.
Apple dropping a $3,500 bomb on the industry has been interesting to watch, but outside of content creators chasing views & techheads, the reception has been lukewarm from the average consumer and Meta has to see that. It's a fine line to walk.
i disagree bc bigscreen beyond had a cheap eye tracking mod thats super small without any heat issues
@@Hallow.X Bigscreen has zero additional onboard compute or really any moving parts at all because the IPD is fixed as well (they set it in the lab when making your facial interface). Easier to add on that mod in that case. Also Bigscreen is $1k WITHOUT any controllers or lighthouses. Well out of Meta’s range.
@@liveopenmike yea ik but the components for the mod cost around 50 dollars and when companies like meta buy in bulk from source the components price goes down buy buy about half
@@Hallow.X That's just the component cost (BOM). You're not factoring in mass production, distribution, and quality assurance at the scale Meta would need to be at. EyeTrackVR, the mod you're referring to, wouldn't work at scale, like a lot of these mods you see for headsets. That's why they never go official. They can't scale properly.
Your typical markup for a product is MUCH higher component cost, so even with Meta's subsidizing a $50 BOM, you're looking at about $100 - $200. Not worth it when the adoption rate is that low.
Just starting. Completely agree about eye tracking off the bat
I have been ripped so hard for that take
@@liveopenmike it’s a good take imo. I think the quest pro shoulda been delayed or somehow worked out using quest 3 chipset also. It really was DOA with those specs and knowledge of quest 3 dropping within 12 months.
Quest 3 was too early for eye tracking to be worthwhile and I think the pro proved that. In the form that meta had it drew too much power and cost too much to include.
I think in addition to tech bringing down the battery depletion and cost, eye-tracking will start to become more important if we start doing varifocal displays, unless the displays somehow deal with it themselves without needing to know the user's eye position/gaze.
Totally agree on being too early. That's one of the main points of my take. I think eye tracking is on all future headsets. I'm really interested in varifocal displays, but (a) I don't think we see them at the consumer for about two generations of hardware (like AVP3, Quest 5 territory) and (b) varifocal, at least in early generations, will likely have the same thermal/power draw issues that eye tracking currently has.
All this REALLY makes me excited for what wearable tech looks like a decade from now though.
reasons for the quest pro "failing" are manifold and don't think any one thing was the cause
No there's never any single reason any project's failure, but lack of software is, IMO, the #1 downfall of that headset. People will shell out money for any product if they see an actual use for it. Quest Pro was/is a decent piece of tech in need of a primary use case.
@@liveopenmike in this case it was a combo of no use case and an aging soc that just made it look like yesterdays jam.
@@onikaze8445 I believe mid hardware can be saved with good software support. Most of us aren't running top of the line PC rigs, but well built, properly optimized software takes a lot of strain off the hardware.
@@liveopenmike mmmm sure but this was an aging soc that was underpowered at release for its use case. It was “better” that the quest one. For them to have pulled it off with the headset specs they would have had to have a vision os v1 level of polish which they just don’t have. My 6mile view says too much inter-dept infighting and no clear leader with veto power.
@@onikaze8445 can definitely agree on lack of vision; it felt lke a gen 1 dev kit
Eye tracking will be must in quest 4 when its more affordable and mature. Foveated rendering is must cuz vr graphics are too demanding. We need mediocre gaming pcs to drive pretty graphics othereise market is too small for serioua development
Agreed. I said in the pod it should be in every Quest headset moving forward. Just didn’t make sense for this one. I will say I’m pretty satisfied with a lot of the Quest 3 upgraded games. I just wish the capture was better for streaming.
Eye tracking for Quest 3? No. You’re right. Not if they wanted the price they were targeting. The real question is why did the Quest Pro launch without the XR2, the Quest 3 lenses, and q3 controller? For games? I couldn’t care less about eye tracking. For system navigation? Absolutely wish we had it…in a better Quest Pro.
I think they wanted to be first to market on a 'productivity' headset knowing Apple was coming soon (turns out they had 18 months to get it right and didn't need to rush). I can't remember who said it, but the phrase "a solution looking for a problem" comes to mind.
The better SoC definitely would've helped, but they were still over a year late with the Microsoft Office integration (and even that was just webapps), Workrooms was... not it when it first launched.
Mid hardware can be saved with a great software library and support. Nintendo's made a living on selling average-power hardware with top tier software. Most of my friends have mid tier PCs, but we're all running (mostly) well optimized software that takes the pressure off the hardware.
Quest Pro launches with the full Google App Store, proper Microsoft Office integration, and a cloud based solution for editing and artistic software like the Adobe suite and I think it's got a shot at being a winner.
For $50 more the psvr2 was able to include eye tracking
@@bxile730 The psvr2 has a full pc in the form of a PS5 doing all of its actual compute work and Sony STILL had to put fresnel lenses in it to get the cost down. This is a valid comparison if PSVR2 were full standalone. Factor in the PS5 and it’s $950.
@@liveopenmike yeap, rn in vr industry none other company can make their prices go that low as zuck. Meta has so much advantage here ı mean they are producing quest 3 for 430$ as far as I can remember. I mean only the cost of producing, we dont even consider things such mass production costs, logistics, marketing etc and yet they are not only ''making vr headsets''. They are publishing the most amount of papers, researchs, ai based models for computer vision and graphics for VR/AR/MR at Reality Labs. Best tech purchase of 2024 for me lol.
@@bxile730 more like $550 more, since psvr2 not a standalone headset and you need a ps5, you depend on sony studios games, $60 additional pcvr dongle cant work with hdr, eyetracking ,haptics or wireless connection, and ofc fresnel lenses.
I looked before I bought it. This will likely be the last one I own. All future ones will be beaming your iris map to the system and all the digital thefts will ramp up. Oh yeah, I'll take that mink stole, here's my "Iris" (software fakeout). Y'all enjoy.
🧐
geez...remove the music in background
There will be a poll this week on the channel. I HATE the sound of my own voice and sadly the second I added the music and some visual cues, these videos started doing better. But I will let the audience decide.
Great talk. No you are not crazy. The Quest 3 is phenomenal as is. It’s understandable why the Quest 3 doesn’t have the eye tracking. It’s all about having a product for the masses for META.
I just wish eye tracking software was being invested in and included on all VR HMDs from last year, and going forward. If the hardware doesn’t have the equipment to take advantage of eye tracking, then developers have no reason to invest in eye tracking and its benefits either.
For instance, mixed reality. Mixed reality is cool, but no game is cool enough to where folks are buying it because of mixed reality. It’s just an added feature in games. And developers are including mixed reality in games since it is a feature. I guarantee you that if the Quest 3 had eye tracking, and was remotely invested in the same as mixed reality or even hand tracking, we’d have games that would wow us due to eye tracking.
As you know, I’m a firm believer in eye tracking (and refuse to buy a PCVR headset without eye tracking) for numerous reasons. Social VR, performance gains, functionality, our eyes being the main sense for VR, it’s cool, it can open up doors to new aspects of VR that we’ve not even began to think as consumers, so on and so forth but until it’s made widely available, devs have no reason and the potential is just postponed.
And like you, I’m super skeptical of that third party solution for Quest 3 eye tracking. It’s been brought up numerous times in my community but I can’t help but to be doubtful. Until there’s a working unit for consumers, it’s just another one of those accessories that will be forgot about in a few months.
Thanks for bringing this up! Here’s to hoping eye tracking is utilized to its potential in the coming years!
See my point about development cycles. These headsets started development probably 2-3 years ago when eye tracking tech was still very nascent and really expensive. I think next cycle of headsets, eye tracking becomes standard across the board.
@@liveopenmike I listened to the entire podcast and understand all the points you made. I can't help but to shake my head when people mention throwing features on these VR HMD's like META put together the Quest 3 in just a few months. The general public and even VR enthusiasts don't truly understand what all goes into making a quality VR headset for consumer release. Years of research & design, with each part meticulously chosen.
But like I said, I understand why the Quest 3 didn't have eye tracking. Unlike the majority, I've been craving widespread eye tracking implementation for years now. It's something that will eventually get here at a large scale and when it does, it'll be a game changer for consumer VR. For now, enthusiasts have their taste.
No they weren't right. It should have been standard by now.
So it's Meta's fault it's not standard at the consumer level yet?
ETDFR wasn't even usable by a lot of devs until URP and Unreal 5 were widely available, so with the timing of devkit availability and devs switching engines, the usage likely wouldn't be any higher than it is now.
Plus I'd like to get more than 90 minutes on my Quest 3. But to each their own. I think it's standard in the next generation of headsets, Quest or otherwise.
@@liveopenmike Eye tracking isn't so Nasa tech. I'll admit that we don't need eye tracking all. it is too simple of a tech to not be a consumer tech.
@@GirlOnAQuest the word I used was nascent, as in still in the early stages
@@liveopenmike lol fair point for the battery problem!
You😜😁❤️
Hello! 😊
@@liveopenmike I meant to write you crazy! 🤭 I like this new format. Good stuff.
nope..
@@captaincrunch72 yup…