I’ve already posted some initial impressions right after I got the 5K+ on Dec 16. In the last two weeks, I managed to spend some more time with the HMD and also hosted a small VR party last night. I wanted to come back and do a proper review based on those experiences.
SOFTWARE… clunky but improving
Initial setup was not straightforward (I had to do the service restart, file deletion game a couple of times till lighthouse tracking was finally working – there should be a „tracking reset” button in Pitool to automate this), but by going through the forums I could get it working within an hour. It was really not clear why you need room setup in Pitool and in SteamVR and which one to do first. This is probably confusing for newcomers.
It’s also unclear when you need to restart which tool. Based on my experience it’s mostly safe to assume that any change in Pitool (FOV, render quality, parallel projections) that affects the rendered image requires a SteamVR restart to take effect.
I’ve upgraded the HMD firmware a number of times, it was pretty straightforward, nothing to write home about there.
I did have some sort of conflict with my Logitech devices: whenever I started Pitool, my Logitech wireless mouse and keyboard would start to stutter and become unresponsive. Then I installed the famous Inateck PCI-E USB expansion card which I bought 2 years ago for my Vive and this solved all of the conflicts. Strongly suggest getting this card for any VR device – the motherboard USB implementations are usually quite messed up.
I know a lot of people complain about the software but it improved a lot in just two weeks (overall stability, added brightness setting, improved tracking), so I’m fairly optimistic about it.
FOV… binoculars to ski goggles
Okay, let’s get this out of the way. In my first review I wasn’t particularly impressed by the FOV although I did mention I had to go without my Vive for 5 months due to moving to a new house. So basically I looked at the Pimax 5K almost as a newcomer and while I recognized this was bigger than the Vive’s, I could still see quite a hefty piece of frame around my view, both on the periphery and the inner nose area.
After spending two more weeks with the headset and also digging out the Vive for a short comparison demo (btw, the Vive went right back into storage), I have this to say: the Pimax FOV is a LOT closer to natural vision than the first generation. You get used to good things easily and going back to the Vive was a big shock, both in terms of FOV and clarity – but more on that later.
To sum it up, the FOV is huge in relative terms (as compared to the Vive), and okay in absolute terms (as compared to human vision).
OPTICS… proper positioning makes a huge difference
If you just hastily put on the headset then a lot of the issues you can hear from others appear: distortions, warping, god rays and eye strain. I assume due to the lens size, proper positioning is more critical with the Pimax than with the Vive or Rift.
Thankfully, there are a number of things you can do to improve the image:
- vertical positioning – this made some big noise on both reddit and the forums and I can confirm it: headset too low and you’ll see warping of horizontal lines to wards the edges. The reason why this was not such a big deal for me is that I already put it high enough (I guess due to my face shape).
- IPD setting – this is trickier. To date I don’t have a surefire way of setting this, the closest I got was watching for god rays: if the IPD is off (too low), I see god rays. Measuring my IPD doesn’t help: even with the Vive I had to set it to 3-4 mm wider than what my optician told me to get the best picture. Based on friends trying it, they get a cross-eyed feeling if it’s set too low, so maybe that’s another way to test this.
- vertical angling: on top of positioning, tightening or loosening the top can make the HMD become angled compared to the plane of your eyes. I found this also makes a difference.
- lens frames: This is more of a QA problem, but worth checking it if you perceive a difference between the two eyes. The frame holding the lens in place can be misaligned (not sure if this is something that happened during shipping) – the frame can be pushed back in place. I made a post about this here (https://community.openmr.ai/t/psa-check-your-lens-frames-if-you-see-discrepance-between-eyes/11942).
The one thing I couldn’t quite fix is chromatic aberration on the edges and I’m not sure if they can do anything about this in software (although I believe they could pre-distort the image for that).
Altogether, I don’t think the distortion is such a huge issue as people make it out to be. Perfectly easy to forget about it when immersed in a game.
PANELS… great clarity, barely acceptable colors:
When net resolution is set to 100%, the clarity is astonishing. You can clearly see distant objects, text is easy to read and overall immersion is greatly improved. There is no perceptible SDE unless you concentrate on it. The infamous black dots (a grid-like pattern of black dots, almost like every 10th pixel in each direction was black) are visible, especially on a skybox or uniform surfaces. During actual game-play though, you don’t see the dots (unless you concentrate on them) but a pattern moving about (like the canvas pattern on an oil painting). This might be annoying while watching movies but completely fine for gaming. For what it’s worth, neither of my friends noticed it or made any mention of it.
Colors are unsaturated and sometimes feel washed out (especially compared to OLED panels). This is again mostly relevant during testing and feels fine while gaming. Not sure about movies, I haven’t really watched anything yet.
Contrast is not impressive, this probably has to do with the black levels and the somewhat lower overall brightness.
And this brings me to the truly disappointing part: black levels are terrible, even using the low brightness level in Pitool. This is most noticeable in Elite Dangerous where most of the screen is occupied by space. I can’t help but cringe every time I leave a station – space is a muddy gray, shadows are washed out. I’m supposed to have an 8K on the way, I’m curious to see the difference, but this is probably my biggest gripe with the 5K.
COMFORT… okay but DAS makes a huge difference
The headset is surprisingly light for its size, even feels lighter than Vive. The stock head-strap does its job but isn’t comfortable in the long run. There is some stitching in the strap that goes over your ear when you pull down the back part to the back of your head (which is the right positioning, as learned from my Vive experience). This puts a lot of pressure on the top of your ears and quickly becomes uncomfortable to the point of being painful.
As mentioned above, proper positioning is key to the best image quality and this is where the stock strap fails in my view.
Due to the nature of the strap, you can and will adjust the side straps independently which can result in lopsided positioning which can be detrimental to image quality. Also, the HMD can move around quite a bit between different sessions (no guarantee to put it on the same way).
Based on this, I decided to give a go to adapting my Vive DAS to the Pimax using some velcro, which turned out to be a huge comfort upgrade. Lot more comfortable and more consistent in positioning. Not to mention the convenience of using a built-in headphone!
The cable is better than original Vive cable, especially for sims – it’s probably very similar to the new slim cable of the Vive (I don’t have one so can’t compare). Feels light, doesn’t tangle too much. One complaint I have is it’s overall shorter than the Vive, mostly due to the lack of a proper breakout box. The one included has built-in Displayport and USB cables so to extend it, you need extender cable which are somewhat harder to find in computer stores than regular ones.
PERFORMANCE… holding back the HMD’s true potential
First of all, here is my rig:
- 8086k @ 5.2GHz HT off
- 2080 Ti @ 1980MHz (stable boost)
- 32GB RAM 3200MHz@CL14
- Aorus Z390 Pro
- M2 SSD 1TB
My primary use case is simulators, most of which are not built for VR but had a VR update added in later in their lifecycle. I play Il-2 Great Battles, rFactor 2, DCS and Elite Dangerous. Neither of these support the angled panels so parallel projection has to be enabled which has a huge performance impact. I value smoothness over quality, so the result is that if I want to have fps anywhere near 90fps, I need to downsample a lot. And here is an important point to make: the only thing that matters is the net render resolution after you set Pitool, SteamVR and in-game supersampling.
Sidenote on supersampling:
Pitool (and most games) use quadratic scale, which means you set how much to upsample in both directions of the view. This results in exponential load on the GPU as 1.5 supersampling really means 1.5 x 1.5 = 2.25 more pixels to render. Most in-game supersampling also works this way (eg. HMD Image Quality in Elite Dangerous). SteamVR uses a linear scale so 150% really means 150% more load. You can check this in SteamVR (or fpsVR) by looking at the final resolution. So here is the formula for net resolution:P^2 * G^2 * S = R
P: Pitool quality slider
G: In-game supersampling
S: SteamVR supersampling %
R: Net supersampling % to be rendered by gameSince what really matters is R, I simply keep both in-game and Pitool settings at default, I set the supersampling in SteamVR on a per app level.
Using above formula, I found I can maintain a reasonably stable 90fps in most simulators (except DCS) with a net 55% supersampling on a normal FOV with parallel projections. Unfortunately using this 55% gives you a much softer and pixelated image, somewhat resembling the original Vive in clarity which is quite disappointing. Basically, for demanding games, you need to make a choice: sharpness or smoothness. You can’t have both on today’s hardware.
For built-to-VR roomscale games, I can disable parallel reprojections and use a net 100% (or more) supersampling for a sharp image. Onward looks great for example (although it has some other issues, like peripheral culling or gray haze).
One important thing to mention: resolution is not consistent across different FOVs as of Pitool 1.95, and enabling Parallel projections support further complicates matters. See my post about this here (https://community.openmr.ai/t/psa-pitool-render-resolutions-are-not-linear/12079). The increased resolution of Normal FOV (PP on) might just be a bug, though.
If SLI could be somehow utilized natively from Pitool without requiring game support and would have significant benefits, I would buy another 2080 Ti to be able to knock back supersampling to 100% (Yes, I have disposable income available for my #1 passion.)
Whenever it goes to reprojection (anything less than 90fps), it feels very much like Vive’s asynchronous reprojection: smooth head movement with ghosting for moving objects. Not bad, but not great either.
As a final note, I really hope Pimax can further work on improving performance, as I think rendering performance is the biggest obstacle right now in the way of more widespread adoption and realizing the headset’s true potential.
TRACKING… feels slightly less stable than Vive
Once tracking is established (which even after the software update is a tad slower than on the Vive), it’s pretty solid.
Sometimes it feels odd, though: things up close would jump around – floating menu panels, cockpit frames in simulators. Not sure if this is a tracking issue or something else.
It seems it’s also easier to block the sensors than on Vive – reaching up to adjust the headset will frequently cause the tracking to vanish (gray screen), even though there is plenty of surface area exposed to the lighthouses. Maybe the placement of the sensors is not ideal. Pimax was touting this as a one lighthouse HMD for a while – not sure if that is realistic with the current setup or if they changed anything in between.
Switching between dominant lighthouses will also frequently pause tracking (momentarily gray flash).
Controllers pair without a problem BUT for some reason cannot be switched off from SteamVR.
I guess Lighthouse tracking is like household work – it only gets noticed when it’s not done. For most of the time, it works great and certainly not something to complain about. Just wanted to mention these issues for the sake of completeness.
ISSUES TO BE SOLVED:
So what are the things that hold this HMD from being truly great? In order of my own priorities:
- Performance: parallel projections destroys performance, more optimizations needed
- Color vibrancy and black levels: hopefully this can be further tweaked in software
- Headstrap: a rigid headstrap (with built-in headphones) is a must – positioning accuracy is critical for a wide FOV headset like this, not to mention comfort.
- Setup and software: simply not ready for mainstream yet – needs to be streamlined with better instructions (something like SteamVR room setup) and the software needs to be further stabilized.
SUBJECTIVE IMPRESSIONS… a big step forward in VR:
This review wouldn’t be complete without sharing my overall impressions from the last two weeks as well as mentioning feedback from those who tried it.
The 5K is basically what the first generation should have been:
• FOV that feels natural
• Clarity that does not hinder tasks like reading instruments or seeing something in the distance
• SDE so reduced that you need to go looking for it to see it.
Spotting and identifying planes in a simulator like Il-2 is a lot easier (even downsampled), spotting hostiles some blocks away is finally possible in Onward and you no longer struggle to read any text in Elite Dangerous. Truly, a dream come true for VR enthusiasts.
