cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
22,062
HTC has announced the all new HTC Vive Pro Eye with foveated rendering and eye tracking. It will blur the details of the screen where the user isn't staring to save system resources. This also allows the image details in the area where the user is staring to be sharper as system resources can be shifted from the blurred area to the detailed, concentration area. It is a professional headset, but the technology is expected to be supported in SteamVR like the HTC Vive Pro is. NVIDIA announced support for the new foveated rendering technology, virtuallink, and eye tracking in the HTC Vive Pro Eye at its CES press conference. Also announced at the event was the new HTC Vive Cosmos. It is a unit that doesn't need external base stations to track the controllers or HMD. It is listed as a premium PC VR system, but it also shows a smartphone in video advertising.

Introducing VIVE Cosmos, a new premium PC VR system, and a vision of the virtual future. It conforms to your wants, your needs with surprising ease and versatility. Whether you're new to VR or familiar, a fantastic voyage awaits. VIVE Cosmos goes beyond everything you could want in one device.
 
+1 for wanting more pixels but hopefully the foveated rendering makes it less important.

It can't physically make the resolution higher on the panel, though.

Realistically, they should have upped the panel resolution as it would have been a perfect companion to foveated rendering.

Also, HTC's support is so bad I doubt I would ever buy a product from them again.
 
Putting aside the predictable "but it's not perfect yet" complaints, this is still progress in this space.

What would really be troubling is if there was no new VR stuff at CES at all.
 
Foveated rendering is actually a big deal. With even a BASIC Foveated rendering algorithm you can DOUBLE performance and take no hit to perceived quality. More advanced, hardware level algorithms can triple performance or more.

Edit: please also note that this is a Vive PRO headset, with 1440x1600 per eye, giving it a huge boost over the OG vive and Rift. Couple that with a need to pump each ~2Mpixel screen at 90Hz, This resolution is pretty much impossible to render on anything other than the top tier of video cards. With Foveated Rendering, that changes.

PiMax was already criticised for being basically useless because the resolution was so high it was practically impossible to drive at the required refresh even with rediculously powerful systems. With the most basic Foveated Rendering techniques, you could go from 45 FPS to 90 FPS, which basically saves the PiMax completely.

The issue is that you need real ground-level support in order to make FR a reality, so by HTC (and valve) essentially spearheading it with this new product, it's going to become much more likely OTHER headsets will support this technique.

Be happy about this.
 
Last edited:
Foveated rendering is actually a big deal. With even a BASIC Foveated rendering algorithm you can DOUBLE performance and take no hit to perceived quality. More advanced, hardware level algorithms can triple performance or more.

Edit: please also note that this is a Vive PRO headset, with 1440x1600 per eye, giving it a huge boost over the OG vive and Rift. Couple that with a need to pump each ~2Mpixel screen at 90Hz, This resolution is pretty much impossible to render on anything other than the top tier of video cards. With Foveated Rendering, that changes.

PiMax was already criticised for being basically useless because the resolution was so high it was practically impossible to drive at the required refresh even with rediculously powerful systems. With the most basic Foveated Rendering techniques, you could go from 45 FPS to 90 FPS, which basically saves the PiMax completely.

The issue is that you need real ground-level support in order to make FR a reality, so by HTC (and valve) essentially spearheading it with this new product, it's going to become much more likely OTHER headsets will support this technique.

Be happy about this.


That’s crazy dude. Thanks for sharing that bit.


Sounds like massive improvement is coming.
 
Foveated rendering is actually a big deal. With even a BASIC Foveated rendering algorithm you can DOUBLE performance and take no hit to perceived quality. More advanced, hardware level algorithms can triple performance or more.

Edit: please also note that this is a Vive PRO headset, with 1440x1600 per eye, giving it a huge boost over the OG vive and Rift. Couple that with a need to pump each ~2Mpixel screen at 90Hz, This resolution is pretty much impossible to render on anything other than the top tier of video cards. With Foveated Rendering, that changes.

PiMax was already criticised for being basically useless because the resolution was so high it was practically impossible to drive at the required refresh even with rediculously powerful systems. With the most basic Foveated Rendering techniques, you could go from 45 FPS to 90 FPS, which basically saves the PiMax completely.

The issue is that you need real ground-level support in order to make FR a reality, so by HTC (and valve) essentially spearheading it with this new product, it's going to become much more likely OTHER headsets will support this technique.

Be happy about this.
Agreed. Specifically eye-tracked foveated rendering is for me one of the pillars of true next gen high end VR. Per my calculations when I was researching foveated rendering tech, with accurate and low latency eye tracking you can actually reduce rendered pixels by a factor of 8 (~12% of full frame) without losing visual fidelity for 110 degree FOV displays. You could do this crudely or with only somewhat accurate tracking and still reduce by 2x or 4x.

The ability to treat gaze as a passive input also has the potential to greatly increase sense of presence (other players avatars eyes moving accurately, NPC’s reacting naturally based on where you are looking etc).

I’m expect for now this relies on developers building this into their rendering pipeline, so will be content dependent. That may be partly why the resolution doesn’t increase over the Pro for now. I wonder if a basic solution could be offered at the driver level... that could really change the game and allow them to bump the resolution to 4K+ per eye level without losing performance.
 
How stupid is HTC, the market is not big enough, the content is not good enough, the product itself isn't cheap enough and lacks many features...
And then instead of improving the headset with features, they segment the market using it by making a new product per new feature or hardware upgrade.
HTC is soooooo bad.
 
So will eye tracking still work with people that wear glasses or will the reflection confuse it
 
Foveated rendering is actually a big deal. With even a BASIC Foveated rendering algorithm you can DOUBLE performance and take no hit to perceived quality. More advanced, hardware level algorithms can triple performance or more.

Edit: please also note that this is a Vive PRO headset, with 1440x1600 per eye, giving it a huge boost over the OG vive and Rift. Couple that with a need to pump each ~2Mpixel screen at 90Hz, This resolution is pretty much impossible to render on anything other than the top tier of video cards. With Foveated Rendering, that changes.

PiMax was already criticised for being basically useless because the resolution was so high it was practically impossible to drive at the required refresh even with rediculously powerful systems. With the most basic Foveated Rendering techniques, you could go from 45 FPS to 90 FPS, which basically saves the PiMax completely.

The issue is that you need real ground-level support in order to make FR a reality, so by HTC (and valve) essentially spearheading it with this new product, it's going to become much more likely OTHER headsets will support this technique.

Be happy about this.
I've used an Oculus Rift before, and the screen door was distracting. Do the Vive PRO variants' higher resolution reduce this meaningfully? I've been waiting on reduced screen door before buying into VR.
 
How stupid is HTC, the market is not big enough, the content is not good enough, the product itself isn't cheap enough and lacks many features...
And then instead of improving the headset with features, they segment the market using it by making a new product per new feature or hardware upgrade.
HTC is soooooo bad.

No doubt they have been stupid, however in this instance I suspect that they’ve just put it out as a development target to get the kinks worked out with the rendering and support.

Once engines can reliably use it and Nvidia support improves then they release a more consumer facing version and homogenise the feature.

If they just did a consumer version then everyone would jump in with a 1060 expecting it to run well, but the software and drivers probably wouldn’t supported the capabilities well and lead to a poor experience (because of crap performance).

Just a guess

Project Cars 3 and DCS 3 on a 4K per eye foveated wireless VR headset. Come on.
 
Well crap. I was gonna buy a rift this month. Now I have to wait and see what the Cosmos does.
 
What a disappointment. For anyone that can afford a Vive Pro, they have a GPU that can run that low resolution just fine without foveated rendering. Completely pointless.
 
How stupid is HTC, the market is not big enough, the content is not good enough, the product itself isn't cheap enough and lacks many features...
And then instead of improving the headset with features, they segment the market using it by making a new product per new feature or hardware upgrade.
HTC is soooooo bad.

This is actually a really good move on their part. This does open the field up to more pcs requiring less expensive hardware to run, which could bring more customers. On top of that, if they had no CES presence this year, people like you would just cry your usual "vr is dead!".

The industry will never be a mainstream thing, no argument there, but it will have to keep releasing new stuff otherwise it's gone way too quickly.
 
If cosmos does have wireless connections to PC, hopefully they learned their lesson with the wireless adapter + Ryzen debacle. and discarded the PCIe req.
 
I just bought a second rift setup for the house cuz the tech is so cheap now.....best bargain out there now, IMHO.
 
I've used an Oculus Rift before, and the screen door was distracting. Do the Vive PRO variants' higher resolution reduce this meaningfully? I've been waiting on reduced screen door before buying into VR.

The effect is still there just much less pronounced. It honestly doesn't bother me at all on the Pro and I can happily game for hours before my face drowns.
 
All right, I admit I'm getting interested, especially the FOViated rendering/eye tracking.

The new Vive Pro Eye - This is essentially the same Vive PRO headset, but just with the FOViated rendering/eye tracking tech right? I wonder if existing Vive PRO owners can buy an upgrade to add the eye tracking or at very least send in their unit for an upgrade? How much new actual hardware tech if necessary? Aside from that, will they be keeping the price the same? The VIVE Pro is at the moment I feel pretty overpriced. For a new user who wants the full kit with controllers and base stations, its around $1100? I felt this was excessive in comparison to the original VIVE even at launch (I thought the VIVE Pro full kit would be the same 800 or 900 that the original VIVE was at its launch ) but now, if we consider that the Vive Pro's specs today (including the ease and lesser expense of the OLED mobile style screens ) it should definitely be cheaper, even with eye tracking. $1100 for the kit or $800 for the visor even after a few year seems excessive, especially if HTC is trying to push the ecosystem adoption. I totally get that it should cost more than the bargain basement VR/AR gear , but when the first generation kit halved-or-beyond in price relatively quickly and started from a lower price despite the narrowness/newness of the original market, things could be improved. Not to mention there are still additional things to purchase (ie the wireless module isn't cheap, the second generation of controllers such as gloves etc... aren't there yet or have to be bought a la carte .)

Vive Cosmos - Nice to see something promising here, but without specs I have to wonder if it is either the new highest end solution in that it doesn't need base stations and is overall a step up in hardware, or if it is basically a lower-end mobile-focused device? The new head mount style and all that looks great; I wonder if it is more advanced than the VIVE Pro in some ways and if those features could be implemented or not . Without specs though and a serious combination between the two, I am unsure in which to be more interested.

I am pleased overall with SteamVR pushing an open (source) specification for FOViated rendering and eye tracking, rather than some proprietary garbage that fragments the market. Things may not move quite as quickly as I'd like, but overall HTC VIVE 's approaches and especially their open partnership with SteamVR and Valve's open source initiatives, makes them the main company in the arena worth supporting.
 
All right, I admit I'm getting interested, especially the FOViated rendering/eye tracking.

The new Vive Pro Eye - This is essentially the same Vive PRO headset, but just with the FOViated rendering/eye tracking tech right? I wonder if existing Vive PRO owners can buy an upgrade to add the eye tracking or at very least send in their unit for an upgrade? How much new actual hardware tech if necessary? Aside from that, will they be keeping the price the same? The VIVE Pro is at the moment I feel pretty overpriced. For a new user who wants the full kit with controllers and base stations, its around $1100? I felt this was excessive in comparison to the original VIVE even at launch (I thought the VIVE Pro full kit would be the same 800 or 900 that the original VIVE was at its launch ) but now, if we consider that the Vive Pro's specs today (including the ease and lesser expense of the OLED mobile style screens ) it should definitely be cheaper, even with eye tracking. $1100 for the kit or $800 for the visor even after a few year seems excessive, especially if HTC is trying to push the ecosystem adoption. I totally get that it should cost more than the bargain basement VR/AR gear , but when the first generation kit halved-or-beyond in price relatively quickly and started from a lower price despite the narrowness/newness of the original market, things could be improved. Not to mention there are still additional things to purchase (ie the wireless module isn't cheap, the second generation of controllers such as gloves etc... aren't there yet or have to be bought a la carte .)

Vive Cosmos - Nice to see something promising here, but without specs I have to wonder if it is either the new highest end solution in that it doesn't need base stations and is overall a step up in hardware, or if it is basically a lower-end mobile-focused device? The new head mount style and all that looks great; I wonder if it is more advanced than the VIVE Pro in some ways and if those features could be implemented or not . Without specs though and a serious combination between the two, I am unsure in which to be more interested.

I am pleased overall with SteamVR pushing an open (source) specification for FOViated rendering and eye tracking, rather than some proprietary garbage that fragments the market. Things may not move quite as quickly as I'd like, but overall HTC VIVE 's approaches and especially their open partnership with SteamVR and Valve's open source initiatives, makes them the main company in the arena worth supporting.

Here ya go! I thought about including this in the original, but I figured someone would be confused and say it is happening. I don't know if they will or won't; but it would be nice if they sold a retrofit kit!
https://www.hardocp.com/news/2017/05/01/htc_vive_getting_eye_tracking_first
 
I am kind of lost here. I though the rendering part is being done by the GPU, not the goggles. I am sure foveated rendering is a big deal but blurring is usually done after a frame has been rendered, no? What current high performance game engine supports this? I remember Carmack talking about this a while ago
 
I am kind of lost here. I though the rendering part is being done by the GPU, not the goggles. I am sure foveated rendering is a big deal but blurring is usually done after a frame has been rendered, no? What current high performance game engine supports this? I remember Carmack talking about this a while ago

Foveated rendering (in it's most simple form) renders the scene out at roughly 1/4 resolution (one half each axis) and then takes the current focal target of the player's eyes and renders that specified area of the screen at full resolution. This means that taken as a whole, the GPU is doing less work by a wide margin.
 
Foveated rendering (in it's most simple form) renders the scene out at roughly 1/4 resolution (one half each axis) and then takes the current focal target of the player's eyes and renders that specified area of the screen at full resolution. This means that taken as a whole, the GPU is doing less work by a wide margin.

OK, so you render 1/4 the resolution and project the produced frame somewhere on the screen. Where does the blur come from?
 
OK, so you render 1/4 the resolution and project the produced frame somewhere on the screen. Where does the blur come from?

From your human vision, more or less. You literally can't resolve things outside the central viewing area. Wondering how much "texture pop" there will be (eye tracking latency). Siting tight with original Vive Pro, haven't had any issues with performance so far.
 
From your human vision, more or less. You literally can't resolve things outside the central viewing area. Wondering how much "texture pop" there will be (eye tracking latency). Siting tight with original Vive Pro, haven't had any issues with performance so far.

I assume it would render past your vision circle as a buffer. All it really would need is to increase that area to accommodate reflex time. If they can make it fluid enough it really is what is needed to help the hardware requirements IMO. I have the regular vive myself and enjoy it.
 
From your human vision, more or less. You literally can't resolve things outside the central viewing area. Wondering how much "texture pop" there will be (eye tracking latency). Siting tight with original Vive Pro, haven't had any issues with performance so far.

Human vision needs something to blur, though. What do they render everywhere but your focus. And how exactly
 
Here ya go! I thought about including this in the original, but I figured someone would be confused and say it is happening. I don't know if they will or won't; but it would be nice if they sold a retrofit kit!
https://www.hardocp.com/news/2017/05/01/htc_vive_getting_eye_tracking_first

Thanks! THat is indeed good to remember. Assuming that HTC 's method is anywhere close to similar in technical methodology or quality, then I will be hopeful the retrofit kit will be easily available! That article mentions how some software is necessary and seems to be limited to/focused on Nvidia GPUs; I am guessing that HTC and especially Valve will not limit themselves in focus to one GPU manufacturer and will bundle the open industry standard into SteamVR etc. These are a couple reasons why its great to have this tech created by or in partnership with a major force in the market rather than a small accessory manufacturer, especially in the case of required software API or whatnot adoption. Thanks!
 
If it will play on 1060/480(Minimum specs) and above then not sure why they would not up the resolution if the eye tracking/foveated rendering will save gpu cycles?
The reason for saving power is to have a higher resolution. Maybe the minimum needs to be raised? Also the pro is 1200? so how much will this cost?
I am waiting for the Rift 2. The Odyssey does look very tempting with the same res as the Vive Pro, but $500.
This will also be new tech and will have to wait and see how it runs/works.

1070/1080 for minimum? Or have different HMD's, a pro(1080ti/2080ti min) and regular.
 
All this foveated rendering for the Vive Pro accomplishes is allowing low end PC's to better handle VR games. That's it.
 
All this foveated rendering for the Vive Pro accomplishes is allowing low end PC's to better handle VR games. That's it.
Well it also allows developers to increase the level of detail for higher end GPU users. So VR games can be more detailed in theory. More than likely game developers will just make games run faster like you said, but there is the option to also make them more detailed.
 
Back
Top