Pimax 5K+ spotted after a 16 month hiberation.... First impressions

[anton_Chez]

Well, it's here. The Pimax 5K+ headset I backed on Kickstarter in October 2017 has finally made it from the depths of hell and everywhere else in between to my house in Australia. What a ride it's been.

To start, the box is fairly nice, nothing too fancy but definitely of Rift quality, nothing flimsy or cheap. The smell is actually like a brand new paid of shoes! I'd say that's the padding. Anyway who cares about the box, hey?

Upon fitting the strap, I was horrified at the comfort. Only after I messed around with it a little bit did I feel I could actually make it work. It's not as intuitive as the Rift but it's workable for me at least. One thing that's immediately struck me is I've nearly got my nose poking through the lenses from the inside out. My face is literally ON the lenses. We are going to have to order custom cushions. It's insane that they've left this unaddressed. I think additional face cushions are a stretch goal reward but I've not see them being any different to the standard one. We need thicker pads, Pimax. I have a small face and a not so overly large nose and I'm struggling. If it turns out my IPD is able to have the lenses moved to the outside of the headset I might just get away with it.

The strap does tighten on your face pretty well. I did not expect that based on other feedback. It's not ski mask type like the Rift but it does have the ability to stick to my face and feel pressure so there's some modular possibilities there especially for larger faces that won't need to tighten the straps as much. Between the side and top strap you should be able to get the HMD sitting nice and flat on your face without any tilting at least while sitting down.

I've yet to power it on so I'll go and do that now. One thing you'll notice is you need an additional power point for this. It comes with an external power source that plugs into the PC side of the cable. It isn't recognised by the PC unless you power it on.

 

[RobertR1]

Thought I'd give a quick update:

- Last night I tried loving the Large FOV which is 170 degrees in AC. I was surprised that there was no performance hit for this that I could detect. The jump from normal to large in notable *BUT* it is not needed. The extra peripheral you gain doesn't necessarily help you drive or see better as the normal fov is more than enough to cover sim racing needs. However, you do get a slightly better sense of speed with the larger fov.

I'd say a 150 degree FOV with higher pixel density/resolution would be the ultimate sim racing VR unit. Going below the normal fov is notable.

 

[anton_Chez]

Steam MIGHT be reducing your SS slider to account for the increase in FOV, keeping the same overall resolution in terms of pixels rendered. Might be why you don't notice a hit. It's 25% more pixels rendered if you were to keep the same image quality, which is quite a bit. If you';re smoothing, it might cover it. If not, I'd say that's what might be happening. I've been looking at the numbers quite a bit.

I can easily race in Small FOV and not immediately notice a difference. It's still MILES better than the Rift even at Small.

 

[Mascot]

What FOV would you say matches your real-life peripheral vision? It's sensing other vehicles pulling alongside that intrigues me most about a wide FOV headset. In the Rift 3/4 of their car is overlapping yours before you see them. Thank God for Crew Chief's spotter to keep me informed..!

 

[RCHeliguy]

What FOV would you say matches your real-life peripheral vision? It's sensing other vehicles pulling alongside that intrigues me most about a wide FOV headset. In the Rift 3/4 of their car is overlapping yours before you see them. Thank God for Crew Chief's spotter to keep me informed..!

What I've read on this subject says that you have 114 degrees of common binocular vision with depth perception, meaning both eyes see the center 114 degrees. Then each eye has additional peripheral vision to either side totaling about 210 degrees total horizontally with another 150 degrees vertically.

The StarVR HMD is the only system out there with about 210 x 130 degrees. Unfortunately it is about $2500 and won't be available until they get sorted.

I should add that in real life a driver is wearing a helmet that does restrict view below that to a value much closer to the restricted view that we get with a Rift. For what it is with I mostly want more FOV for room scale games. If I was just sim racing, I'd probably just want more resolution like many do.

 

[RobertR1]

What FOV would you say matches your real-life peripheral vision? It's sensing other vehicles pulling alongside that intrigues me most about a wide FOV headset. In the Rift 3/4 of their car is overlapping yours before you see them. Thank God for Crew Chief's spotter to keep me informed..!

Snell helmet certification requires 180 degress of horizontal FOV. Page 18 https://www.smf.org/standards/sa/2015/SA2015Final3252014.php#_CONSTRUCTION

Your eyes when looking straight ahead have an optimal viewing fov of 110 degrees but a peripheral view of 210 degrees. Look straight ahead. Everything in that 110 degree cone will be very clear. Now looking straight ahead bring up a hand to the side of your head. You can easily detect an object there but it’s not detailed. You’d have to turn your head and bring it into the 110 degree range to get maximum clarity.

The Pimax large is 170 degrees give or take depending on the distance to your eyes. What this means is that just as you’re able to see your hand in the experiment above, you’d see a car creeping up alongside. This would be without having to get it into the range of the 110 degrees of the rift and other displays. The same car on the rift and others would require you to turn your head or wait until the car is forward enough to fit into the cone.

The extra fov also give you a wider view of the landscape which really adds to the immersion and sense of speed which isn’t often mentioned.

 

[Mascot]

Snell helmet certification requires 180 degress of horizontal FOV. Page 18 https://www.smf.org/standards/sa/2015/SA2015Final3252014.php#_CONSTRUCTION

Your eyes when looking straight ahead have an optimal viewing fov of 110 degrees but a peripheral view of 210 degrees. Look straight ahead. Everything in that 110 degree cone will be very clear. Now looking straight ahead bring up a hand to the side of your head. You can easily detect an object there but it’s not detailed. You’d have to turn your head and bring it into the 110 degree range to get maximum clarity.

The Pimax large is 170 degrees give or take depending on the distance to your eyes. What this means is that just as you’re able to see your hand in the experiment above, you’d see a car creeping up alongside. This would be without having to get it into the range of the 110 degrees of the rift and other displays. The same car on the rift and others would require you to turn your head or wait until the car is forward enough to fit into the cone.

The extra fov also give you a wider view of the landscape which really adds to the immersion and sense of speed which isn’t often mentioned.

And that's why foveated rendering will be such a boon when it finally lands. There's really no reason at all for the image to be at full resolution around the edges of the HMD display. Even if you move your eyes to the edges of your eye sockets the image blurs in real life, and it's the same in an HMD. You move your head instead wherever possible to keep the image centralised and clear and also to reduce eye muscle strain. Peripheral vision is a blur in real life so there's no need for it to have full resolution in an HMD, it's just using unnecessary resources.

 

[Steve D]

After having an epic, close race at Ibarra in the MCR 2000 last night, (Rf2), with the Rift strapped to my bonce, I definitely want a bit more FOV! Lots of traffic, positions being swapped continuously and three abreast on plenty of occasions would have ended in tears had it not been for Crew Chief. Still have a bloody stiff neck this morning!

 

[Mascot]

Slightly off-topic but not common knowledge: in AC (content manager - shader patch) 'Neck' (Ilja's version of Real Head Motion) is now enabled for VR.

 

[RobertR1]

And that's why foveated rendering will be such a boon when it finally lands. There's really no reason at all for the image to be at full resolution around the edges of the HMD display. Even if you move your eyes to the edges of your eye sockets the image blurs in real life, and it's the same in an HMD. You move your head instead wherever possible to keep the image centralised and clear and also to reduce eye muscle strain. Peripheral vision is a blur in real life so there's no need for it to have full resolution in an HMD, it's just using unnecessary resources.

You would want great eye tracking to go with foveated rendering. Depending on the turn, you don't always need to turn your head to spot your brake marker, turn in point, apex and exit but you do need to look at these points on the track. With a fixed fov that would be quite unpleasant and you'd always be drifting between the sweet spot and blur. Eye tracking would solve that issue by moving the "clarity cone" to where your eyes are focusing.

 

[RCHeliguy]

You would want great eye tracking to go with foveated rendering. Depending on the turn, you don't always need to turn your head to spot your brake marker, turn in point, apex and exit but you do need to look at these points on the track. With a fixed fov that would be quite unpleasant and you'd always be drifting between the sweet spot and blur. Eye tracking would solve that issue by moving the "clarity cone" to where your eyes are focusing.

I agree that foveated rendering requires good eye tracking.

However I did read recently about an HMD manufacturer that was planning a short term solution similar to what Mascot mentioned which uses a lot less processing. They are planning to put a small high resolution display permanently centered in front of each eye surrounded by a lower resolution panel. I don't remember what it was called or have a link, but it makes some sense. It allows for a natural appearance IF you move your head and gives you the peripheral cues that would cause you to focus and move your head to look at somewhere.

It might be a reasonable stop gap until eye tracking is refined.

 

[anton_Chez]

Pimax had a stretch goal relating to eye tracking but I am unsure what this actually refers to. They also have a version of foveated rendering but Robert has said it's not great on the 20 series so far. I'd like to try it so it will be good to have it come to the 10 series shortly.

Overall I am extremely impressed and satisfied with my decision to back. The road has been long and frustrating but at the end of the day that's behind us now and I hope more people give it a look in.

 

[RaceNut]

I'm just glad to see more innovation and on-going development in VR. The fact that Oculus has sub-contracted the manufacturing of the Rift-S (providing a minimal incremental visual upgrade) suggests that this segment of the market is certainly not their primary focus so it seems that others have to pick up where oculus is dropping the ball, in developments that can benefit this market segment - at least.

The new HP Reverb looks interesting, and at a bit lower price point than other HMD's. It's limited in tracking capabilities but, should be sufficient for Sim-racing duties.

Still, it seems that we are - at least, 5-years away from having a truly next-gen VR HMD that can achieve a combination of significantly better image quality combined with a FOV closer to reality, and one that doesn't require a NASA-spec PC to use it.

 

[RCHeliguy]

I think creating 20Mp HMD's will be relatively easy in a couple years. StarVR has proved we can create the optics to give you a very undistorted view of 210 x 130 degrees FOV and eye tracking for $2500 but not released yet.. We have a new VR-link standard for a very simple connection ( no on is using it yet ). I don't think the headsets possible are the blocking factor.

Your "NASA spec PC" is the limiting factor in getting to HMD nirvana. Even with eye tracking, there is additional processing required.

 

[Deleted member 197115]

I'm just glad to see more innovation and on-going development in VR. The fact that Oculus has sub-contracted the manufacturing of the Rift-S (providing a minimal incremental visual upgrade) suggests that this segment of the market is certainly not their primary focus

That's a very good "guess", they are essentially dropping out of PC market for now.
https://www.roadtovr.com/facebook-jason-rubin-oculus-interview-gdc-2019/

Beyond any shadow of a doubt, at some point we will have a next generation [headset] where we add some sort of feature that breaks all the old stuff and makes it either not work or seem obsolete,” he said. But the company presently believes that growing a cohesive audience is more important than pushing technical boundaries.

Why would Iribe leave the company otherwise.

I have high hopes for Valve Index (preorders starts May 1st), although silence on the actual spec is quite suspicious.

 

[RCHeliguy]

I have high hopes for Valve Index (preorders starts May 1st), although silence on the actual spec is quite suspicious.

Fingers crossed! The perfect tracking and state of the art hand controllers would be great!

I'm not expecting the HMD to be earth shattering. However if it actually is 140 degrees horizontal FOV that would definitely get my attention.

 

[RCHeliguy]

The Valve Index controllers ( aka "Knuckles") and base stations are compatible with the Pimax. So getting back on topic. For those of you with a Pimax looking play in room scale, these Valve Index controllers should be a HUGE step up from the Vive controllers or likely whatever Pimax is working on.

BTW there will be some rewriting of some games to support the individual finger tracking the Knuckles allow.

 

[Deleted member 197115]

Fingers crossed! The perfect tracking and state of the art hand controllers would be great!

I'm not expecting the HMD to be earth shattering. However if it actually is 140 degrees horizontal FOV that would definitely get my attention.

I wouldn't hope for that, it is most likely LCD version of Vive Pro, same resolution. They won't push FOV that high with listed minimum requirements. FOV is what kills performance, not resolution, as you have much more of the scene to render, Pimax proves this perfectly. Resolution is just GPU which is rarely a bottleneck for single rendering thread sims.

 

[Emery]

I should add that in real life a driver is wearing a helmet that does restrict view below that to a value much closer to the restricted view that we get with a Rift.

[edit: I see someone beat me to the punch]
Nope. Racing helmets are required to have minimum 180 degree field of view horizontally. See the Snell Foundation documents.

Wish this idea that racing helmets have low FOV would die because it is simply not true. Alas, we're on the internet and people keep repeating alternate truths because they find old posts that back their viewpoint.

 

[RCHeliguy]

[edit: I see someone beat me to the punch]
Nope. Racing helmets are required to have minimum 180 degree field of view horizontally. See the Snell Foundation documents.

Wish this idea that racing helmets have low FOV would die because it is simply not true. Alas, we're on the internet and people keep repeating alternate truths because they find old posts that back their viewpoint.

Thanks for that. Believe it or not I searched on FOV for racing helmets and didn't find squat. The racing helmets manufacturers I found would talk about having superior FOV and then not mention what that was. I was actually curious.

 

[Deleted member 197115]

Worth to mention that Snell certification is voluntary, not a requirement, and not adopted by all manufacturers
https://www.smf.org/helmetfaq#aWhySnellHelmet

"So, why aren't all helmets certified by Snell?"
Snell Standards are voluntary. Some helmet manufacturers do not believe that they need Snell to demonstrate that their products are among the best protective headgear. Others believe that they need only to produce helmets that meet Government or consensus helmet standards and some try and just can not make the grade. A very few helmet makers are truly inconsiderate of the consequences of making an inferior safety product (they are not usually around very long).

As well as Snell itself allows helmets with lower peripheral vision.

Some competitive applications may require helmets with more restricted visual fields. When justified, special addenda to this Standard will define reduced visual fields, the procedures for determining whether a helmet satisfies the requirement and the additional labeling requirements warning that the headgear may be appropriate only for certain uses.

And Snell is not the only standard, plus as mentioned, completely voluntary, non mandatory requirements from private, non profit organization.
The regulated and required ones are DOT in US and ECC in Europe.
And this is DOT requirements
https://ultimatemotorcycling.com/20...lmet-standards-explained-dot-ece-22-05-snell/

The standard also requires peripheral vision to be not less than 105° from the helmet midline. Projections from the surface of the helmet (snaps, rivets, etc.) may not exceed 5 mm.

Same for ECE

The ECE standard, which is accepted in 47 countries, is similar to the DOT standard in several ways, for example: like the DOT standard, peripheral vision through an arc of 105° from the helmet midline is required.

Here goes your "busted" myth, our current gen HMDs FOV exceed that already.