Tech Talk #17 : Cambria

hmm, you’re still aware that the Cambria isn’t really designed for the game fraction?
To me, this sounds more like a work space, which requires far less performance.

Yeah it’s definitely not aimmed at gamers, although I guess many will still view it as a potential new headset through those lenses as a lot of us are always wanting the next best thing.

Personally, I am currently working from 4 different locations a week due to life circumstances out of my control, so cambria is actually quite appealing as I’d like to have a headset where I could have the same virtual office space and remote desktop into my workstation remotely and work like that with the same monitor setup as I physically have at my home. That said, I’m not convinced that the PPD of the cambria will enable this, although I think the comfort will, but working in code I feel like 30-40PPD or so is likely the minimum.

1 Like

I will most likely get the Cambria as a roomscale alternative to my VR-3 whcih is perfect for my sim rig, but too heavy and tethered for roomscale experiences.

I personally think that both the Cambria as well as the Apple VR HMD will make great use of passthrough AR; I tested a Varjo XR-3 and was extremely impressed with the quality of its passthrough. Using high quality passthrough can also offer true Mixed Reality content, such as showing hands and e.g. your real life steering wheel or cockpit within a sim, which as a simracer would be an amazing increased of immersion.

3 Likes

I am just looking at the specs & capabilities. The Quest 2 wasn’t focused on providing PC VR either yet does a decent job at doing so in a wireless manner.
The lack of a bump in resolution is indeed a question mark I have too but the features I mentioned make me seriously consider it.

1 Like

There is a bump in resolution 1920 x 1832 → 2160 x 2160, plus I would imagine more of that resolution on Cambria is being actually used due to seperate panels (plus the new orientation giving a bit more vFOV).

The bump isn’t all that significant though (about 15%) but I would imagine in practice it equates to more like 20% difference. If wifi 6e facilitates near lossless compression and low latency (like resported on the index 6e wireless adapter) then that would actually be fairly compelling, especially combined with the form factor, self tracking controllers and imrpoved colours provided by the miniLED.

1 Like

I’m not sure why anyone is assuming the Cambria is going to have this technology being developed by another independent company which is not owned by Meta.

It seems more likely to me that Cambria will release with essentially the same wired and wireless PC link feature that the Quest 2 has now. The hardware may not even be capable of implementing Nofio’s solution. Based on its specifications, I think it’s likely to be using customized hardware acceleration.

Think we might be misunderstanding eachother here. Afaik the index wireless adapter isn’t anything massively different from Airlink or Virtual Desktop, it just interfaces with a headset that isn’t originally wireless and using wifi 6E instead of just wifi 6. It seems like Wifi 6E is the major aspect of what enables them to get essentially lossless video and super low latency. There’s no reason to assume that Meta can’t do the same with 6E on the Cambria, perhaps it might require a dedicated dongle or base station of sorts rather than just a 6E enabled router.

I highly doubt it is using customized hardware acceleration, it will almost certainly be using off the shelf components and the meat is in the software. A product this niche from a company this small can’t really afford to be developing custom chips and acceleration hardware, but I could be totally wrong.

From what I’ve heard so far, it really does seem to hinge on wifi 6E (I’ve also been talking to someone that used it and spoke with the team in person and they are of the same opinion that 6E is the magic sauce here).

1 Like

I don’t think that’s actually true. Quest 2 is running its wireless connection through the XR2. Whereas I think the Nofio is running dedicated hardware which outputs a DisplayPort signal which is able to achieve much higher pixel rate and lower latency. According to the recent Thrillseeker video, it was even able to run 144Hz with no detectable latency or display artifacts. I don’t think the XR2 would be able to do the same thing in software.

You can see this sort of thing in the diagrams for wireless support on the Pimax 12K. The 60GHz support does not route through the XR2. It uses specialized hardware for a direct path instead. The 12K’s WiFi 6E support does route through the XR2, but I think this is something more like Airlink/Virtual Desktop and will have more latency than the other path (though presumably still less than if it was just using 6).

In the Thrillseeker video, it is said to be using a custom video codec in order to achieve this low latency. I agree that 6E will be a part of the “magic sauce”. But the custom codec is likely to be an even bigger part of it.

Video codecs like h.264/h265 are not designed for such extremely low latency real time. They’re more focused on compressing stuff like movies. Especially when you get into deltas from future frames (B frames). The video codec itself produces substantial lag even if network transmission is instantaneous. I’m pretty sure the latency they were claiming is below what’s possible with just the video codec itself with no network transmission at all.

I agree with you that Nofio is likely too small to be working with custom silicon. But hardware acceleration generally only supports major codecs. So they’re probably using a chip where they’ve changed the microcode to speak their custom codec.

According to the video, Nofio does intend to license this technology to other manufacturers for use in their headsets. But I expect that it’s not just a software thing that can be run on an XR2 or anything like that. It’s probably more like a reference hardware design, firmware, and software. Something that needs to be built in.

1 Like

I don’t see why it’s a stretch to think that a company such as Meta wouldn’t be able to utilize 6E to the same extent using a low latency codec. I’ve seen nothing to suggest that any of the hardware in Nofia is unique or proprietary, it seems like it all comes down to software. I don’t think they are big enough to develop a proper custom hardware decoder. It’s totally possible that Cambria doesn’t include anything to take full advantage of Wifi 6E, but that’s just one headset. My point is that 6E is very exciting because it seems to enable wireless streaming of VR that I didn’t think would be possible over wifi until wifi 7, I assumed we’d need to rely on wigig for the next few years for good quality wireless VR.

It definitely could be reference hardware design etc like you said, but nothing that other companies couldn’t do themselves imo. Wifi 6E itself is an add on to the XR2 afaik, like a small add on board or such, there’s no reason to think that just because it’s the XR2 that there couldn’t be additional stuff added on to facilitate similar lossless low latency streaming over wifi 6E.

1 Like

It’s certainly technically possible that Meta could be doing the same thing as Nofio already. But my point is that such a thing is not part of any of the rumors or unofficial specifications about Cambria at all. It seems to be just this assumption that because this startup Nofio is doing it, then Meta is probably also doing it, too. And I see no reason to assume that at all.

In fact, I doubt that Cambria will even be relying on PCs for its “productivity”. I think Meta wants to go full standalone with it, and so like on the Quest 2, the PC link will remain something of an afterthought. Meta may not even be interested in Nofio-like technology. This is why I think the airlink functionality on Cambria will essentially just be forwarded from the Quest 2 as is.

I think Meta’s primary plan is to have a bunch of native productivity apps on the Cambria. Meta’s whole trajectory up to this point has been trying to encourage customers off of PC and onto their native platform, and I don’t think that’s going to change with Cambria.

1 Like

Good points, maybe I came across as focused on Cambria too much (although logical cause it’s a cambria thread). I’m mostly excited about 6E in general, for the reasons I mentioned earlier (I thought wigig was the only real high quality option for now, and the line of sight aspect is a borderline deal breaker for me for wigig). So I’m a lot more optimistic about nearterm wireless streaming thanka to Nofio that I was previously.

We also don’t know for certain if it does require any specific hardware. I do agree that it’s unlikely that Cambria will push it much although I think Cambria will actually be relatively PC focused for productivity (against Meta’s will nearly) because Meta just doesn’t have a useful software stack for standalone productivity yet. I guess you could claim cloud apps will bridge that gap but from Meta’s perspective that’s basically the same as 3rd party apps.

I do think they want standalone productivity to be their be all end all but their productivity software just isn’t there yet. They might release some with Cambria but I would eat a small family alive if it was actually up to snuff for any real productivity.

Anyway, that’s all beside the point. I agree that 6E streaming on Cambria almost certainly won’t be up to the stand of Nofio, I do think it will be an improvement though and I’m excited about 6E in general, even for Pimax reality series etc.

1 Like

Although the resolution of the panels indeed is a bit higher, unfortunately Meta is apparently rendering the exactly same resolution as for the Quest 2, probably to avoid a) update requirements for the existing app base, and b) performance losses b/c the additional sensors may create a greater side load on the XR2 as compared to the Quest 2 sensors (but then again the Cambria XR2 iteration is said to perform 20-30% better).

Let’s see, it may or may not pan out as a good visual experience with all the advantages of wireless, high comfort & small form factor. But I am definitely very curious to see what the result of all the efforts which went into the Cambria development are!

1 Like

The supposed default render resolution of Cambria is the same as Q2 panels but it’s higher than the default render resolution of Q2, pretty much no apps on Q2 render at the panel resolution.

Plus higher res panels will look better anyway, even at the same rendering resolution. I’m mostly interested in PCVR personally so if I was using Cambria I would be driving the full resolution most of the time.

Not sure if the rendering resolution on cambria standalone is for compatibility or just because it’s still the XR2 just potentially an updated version with better cooling so performance won’t be that much better.
It does make sense that it would be compatibility seeing as they landed at that very specific number.

I might be wrong about all that though, just going off stuff I remember from the top of my head.

1 Like

I wonder what makes you and some others so confident in this - from my experience in the past 20-30 years native resolution generally is the best, unless you have proper ratios like ideally 4:1 more resolution. But 15% more resolution wouldn’t seem to help, rather potentially decrease the visible quality depending on how smart the upscaling algo’s are. (I am assuming here that we are not talking about SDE, because that is no longer a greater concern anyhow)

Am I missing something ?

1 Like

Because you’re thinking of pancake monitors, not VR displays. The higher the resolution the better the clarity, even with lower rendered resolution (obviously at some point the lower the resolution it stops mattering how high the resolution of the panel), SDE still matters too (for some G2 level is enough but for others it’s still not enough, VR is extremely subjective). Also if you’re rendering 1:1 on a vr panel, it’s not going to be 1:1 like a traditional monitor due to barrel distortion and other details (discussed well here and explained better by other people than me: New to VR and Pimax. Lost and confused - #3 by jojon )

What you’re saying is true for traditional monitors but displaying an image on a VR headset is a whole different ballpark.

That said, for the cambria case that we are talking about there probably would be fairly little difference, but we’re also talking about standalone. My main concern is with PCVR so the panel would be run at full res anyway.

2 Likes

Exactly.

The specific reason why is the scaling post process step. Whenever you perform a scaling operation on bitmaps, you are losing fidelity because you’re mapping pixels from one grid onto misaligned pixels on another grid. This can only ever be an approximation. That approximation gets better by oversampling so that you’re losing less of the original information during the translation.

On a conventional monitor, rendering at exactly native resolution avoids this post process step altogether. So it avoids the quality loss associated with this step. And it avoids the computational cost, too. So on a conventional monitor, there are strong reasons to favor rendering at exactly 1:1 with the physical panel (or at least an exact multiple of the physical panel resolution so that the grids are not misaligned during scaling which reduces the quality loss). These effects mean that you want to avoid buying a panel which has higher physical resolution than your GPU can handle at 1:1. Higher resolution panels can actually be detrimental to overall performance.

None of that applies to VR displays because the scaling post process step is present no matter what. The reason why is because the image always needs to be distorted to counteract lens distortions and chromatic aberration. Scaling and distortions are applied together during this same post process step so that you’re only suffering the fidelity loss of translating from one bitmap to another bitmap one time per frame.

That is to say that performing scaling at the same time doesn’t make the quality degradation any worse. It’s essentially a zero cost operation while you’re having to distort the whole image anyway.

It is necessary to unlearn the wisdom we’ve learned from decades of experience with conventional monitors which are changed by this. There are no special resolutions in VR. There is nothing special about 1:1 native panel resolution or getting at least close to it or anything like that. All of those effects do not apply anymore.

Given that we can’t avoid this post process step in VR, then there are two ways to reduce the quality loss. One way is increasing the resolution of the input bitmap (ie. oversampling). In that case, the pixels are still misaligned with the output bitmap pixels, but they are smaller and there are more of them, so the approximation will be more accurate.

The other way is increasing the resolution of the output bitmap which helps for similar reasons. The more and smaller pixels you have to work with on either the input or output bitmaps, the more accurate the approximation becomes. In fact, if your output bitmap could have infinite resolution, then there would be no quality loss from the scaling operation.

This is why the 6K resolution panels on the Pimax 12K will inherently produce a higher quality image regardless of the input resolution. The closer you get to that infinite output resolution, the better. From the display quality and performance perspective, there are no inherent downsides to the higher resolution panel like there would be for a conventional monitor.

Similarly with the Cambria, a small increase in physical resolution of the panels is always going to produce a small improvement in display quality, not worsen it in any case like can happen with a conventional monitor when you go only a little above 1:1 native resolution. Since there are no special resolutions in VR, increasing either the rendering resolution or the physical panel resolution always produces better quality in all cases.

1 Like

Ok, fair enough, I had forgotten about the barrel distortion effect so it will probably help a bit.

Let‘s see, if that is noticeable in the Cambria.

Its confirmed to release this october 2022 !

1 Like

Nice! Hopefully review models go out early. I have gotten every Oculus headset ever made, have them on display actually haha, but I think seeing as this is a “meta” headset and not that appealing to me in general then I’ll be giving this one a miss. I think my days with meta are over unless this ends up really being something way better than we had expected with properly good software. I do want a headset as a monitor replacement device but the PPD on this is going to be too low (unless the leaks are wrong) so likely will be skipping this headset entirely.

Still, very interested to see reviews and possibly test it etc. The comfort is likely the most appealing aspect of it.

1 Like

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.