Something to mention that i don’t really see anywhere and that is causing confusion. Pimax should be posting this in a big sticky as its completely opposite to what all the other manufacturers are doing.
IPD wheel on the HMD will give you a ipd number but it only moves the lenses. It does not have any effect on the image. Seems to be optimized for an ipd around 64-65
IPD slider on Pitool will adjust stereo separation and move the image
Then you have the actual center of the lenses being about 4-5mm off that the actual number so if you are under 63 IPD you probably wont get them centered to your pupils as the minimum possible measures around 63-64. Adding to that im 64 and im slightly off so it sounds about right. 60 = 63-64 center rings on the lens.
Now there is some talk about the lenses and the canted nature VS sweet spot and such that im not knowledgeable about but id rather have 2-3mm extra to get a sharp image and both eyes in the sweetspot than 5 degree more fov. I dont notice any change in 3d geometry with the lenses in any position either. With all the fiddling you have to do anyway and given people have modded this i think they could possibly include it.
EDIT:
I apologize, checking with the lens off, there is no doubt the image is moving with the mechanical IPD wheel
the recap looks valid, the problem is that all those stuff is really subjective & depends on too many params, as other mentioned they loss 3d geometry & 3d depth sense if go with lower IPD while trying to fix it with software IPD slider, but for me it also works (Im about 68ipd) but still use min mechanical IPD to get both eyes in focus, but some reason 5kXR works better than 8k & 5k+ but those also work quite good, I guess it’s coz XR has better lenses revision or just manufacturing variance.
Afaik when you move lenses the software automatically moves image on panels (they loss SDE coz screens are too big & they need to have margin for IPD adjustments) to be at the same spot as before against lenses axis, so you can go from min to max IPD & it works, I believe it’s the same for mainstream devices there is only difference it’s not canted so optic works a bit differently & you quickly loss sharpness on e.g. CV1 if you go out of your IPD, while on canted if can get decent picture on more IPD values, coz there as you mentioned 1 eye in focus at once problem but it doesn’t block to perceive image sharp if your brain trained to do so (there were reports that glass wears have better results with it)
I dont know but I guess the “GPUpower” and “GPUhorsepower” is set for some specifik GPU’s ?
With my 2080ti, the values was below and may have been benifiting from the incerased value but with my 3090 I get ““gpuSpeed0” : 2728” and so on. GPUhorsepower around 2700 as well.
Setting 2000 feels like tuning it down and not taking all the GPU power into consideration?
Just got the 3090, and also just got the 8KX so I have no experience from tuning Pimax. (have 4 years of VR with other brands though.)
GPUSPEED and GPUHORSEPOWER are only used to determine the resolution of SteamVR’s menu screens and overlays (e.g. background); they do not affect PiTool’s RenderQuality or SteamVR’s SS.
No, I dont have bad image quality like that*. Didnt know what the GPUpower was used for.
Just changed the GPU and havent even tried the 8KX with the new GPU yet.
(* The 8KX looked blurred to me the first tries, but that’s sorted out as Pimax having a shorter focal distance than my aging eyes can do. Glasses fixed this.)
Apparently if your default values are bigger you don’t need to tweak it, it was written when 3000X weren’t available to market, but the principle is the same, if steam menus & overal quaility is ok then keep it intact
Just want to add my experience with receiving my two 8kXes:
Both need completely different ipd-wheel and soft-ipd settings to give me the same sharp image.
So whatever someones setting is, it is very likely that it does not work for anyone else. I don‘t know what the reason is, i can only guess that it is the non uniform assembly of the screens/lenses. Or maybe because the ipd wheel is to wobbly. Who knows.
I was really shocked when trying my second 8kX and the image was a complete blurry mess. Then i did the trick with adjusting for one eye and then for the other. Then adjusting the ipd wheel accordingly and afterwards adjusting the software ipd for the difference to my measured ipd.
Not yet.
You mean that if i max and min the ipd wheel then afterwards the nice and crisp image i have right now would change (or at least the reported ipd number)?
That’s what I thought but no the lenses in mine where pressed so hard against what ever part they are supposed to just gently glide over, that correct IPD adjustment was impossible. I could do 2 full rotations of the IPD wheel where the lenses moved but the sensor for the software adjustment did not pick up the movement. LH tracking was a giant mess with it too.
It was a real lemon how it pass QC is beyond my comprehension.
That specific 8Kx has been sent off for RMA replacement.
My 5K+ (with it’s misaligned optics) works flawlessly compared to the 8Kx I received.
First of all, Thanks for this detailed post, a Few days ago I received my Pimax 8K X so looking for how to get the best of best from my HMD however I am happy with the current image quality but have felt like it can be clearer than what I see now and get the perfect balance between Picture quality and performance. I want to know all those configs edits, bat files, etc. you suggested are still needed to do it manually after Pimax Experience software? Thanks
From my understanding PE should do it on it’s own, I just briefly checked it & prefer plain pitool for my liking, so not sure if PE works as expected, last time I checked it had parallel settigns with pitool, e.g. if you change it through pitool you have one settings if you do it with PE you have another ones, also separate colours channels constantly got switched to global channel settings in PE so I stopped use it.
Maybe folks know if steamvr settings adjusted correctly nowadays & it’s not required to do it manually, Im almost sure its automatical there as it just was the main point of that software.
Easy way to test - if in steamvr (in title that supports on the fly render resolution (super sampling or render quality) change e.g. DCS doesn’t support it but VRChat does) if you increase your SS higher & performance gets worse gradually then limit setting is in place & steamvr doesn’t limit your max resolution, if perfomance isn’t changing when you increase resolution in steamvr then maxreccommended setting is wrong & you need to adjust it, restart steamvr & check again.
There is point once picture quality stops improving & image looks the same, so no need go higher than that, you also can set potato (small) FOV & test maximum supersampling rate your system can handle if it starts lag a lot to see how it could look with that settings. In comparison on mainsream HMDs like Vive Pro or Valve Index if I run VRChat or DCS it doesn’t look nice with 100% steamvr SS, but once I run it at 350% or higher it really shows the potential of the screen, though in DCS I have to use smartsmoothing & I have 5950X & 3090, I have 3 Pimaxes & recently tested them all again (OG 8k, 5kXR & 5k+) from these I found 5k+ is best one still & what I really like about pimax is FOV,
I bought 1 week ago 8KX as couldn’t resist (trying to do it since Autumn & almost 4 month later decided to buy it still lol) So Im not yet sure how much SS that HMD needs to look the best, I guess it should be bigger than 100% coz chromatic abberation software tricks & some other stuff lower perceived resolution in HMDs
As you’re ok with image I guess you already get quite ok result & main problems you may have can be fixed with optics calibration, trying different distances for eye relief with different cushions etc.
but no the lenses in mine where pressed so hard against what ever part they are supposed to just gently glide over, that correct IPD adjustment was impossible. I could do 2 full rotations of the IPD wheel where the lenses moved but the sensor for the software adjustment did not pick up the movement. LH tracking was a giant mess with it too.
how it pass QC is beyond my comprehension.
