Pimax Crystal Super

[CAG_Sanders]

Update on the tariffs from Pimax. My order is waiting to clear customs in the US right now.  https://pimax.com/blogs/blogs/pimaxs-measures-to-mitigate-the-impact-of-the-us-tariffs

[Varibraun]

4 hours ago, CAG_Sanders said:

Update on the tariffs from Pimax. My order is waiting to clear customs in the US right now.  https://pimax.com/blogs/blogs/pimaxs-measures-to-mitigate-the-impact-of-the-us-tariffs

 

That seems like good news all the way around with some thoughtful planning and a customer friendly approach by Pimax.  It doesn't seem that my preorder hit the first batch since I haven't heard anything since 4/4, so I look forward to hearing your thoughts once yours lands at your doorstep.

[von_Tom]

 

Dunno about it.  Historic quality control issues are a turn off as is the fact that current hardware won't be able to run it at a decent fps.  Less than 90fps doesn't work for me.

 

I'm looking forward to seeing actual user thoughts.

 

von Tom

Edited April 11 by von_Tom

[chiliwili69]

Wow, this is really what I feared. The ratio rendered/physical is along the line of previous Pimax headsets:

 

 

You need to render 2.66 times the pixels of the physical panel. So, this is full waste of GPU power spent in this poor geometry of aspheriical lenses.

 

It means that even with a 5090 you only get 45 fps!!

 

 

I had some hope in the device which is the only one giving good FOVs and DP cable.

All the other solutions (BSB2, Meganex 8K ) are MicroOLED based with small FOVs or has not DP cable.

 

I have no problem in getting a 5090, but no, even with that you are at 45fps in IL-2.

 

That´s really bad.

 

Let´s see what default resolutions we will have with the Air Dream and that new pancake lenses (but forget this FOVs)

 

[zeppelinzach]

thanks chiliwili, your posts about choosing HMDs are always very helpful.

 

next time i get a new hmd i'll check for the resolution graphs you post.  this is essential info that the popular youtube-style reviewers dont bother to test!  it should even be on sites like vr-compare.

 

prices of HMDs are much higher than a few years ago ($400 G2 vs $2000 Somnium) but what has really improved?

  even some fresnel lens HMDs from before 2020 have better clarity than new models.

  only a few new HMDs have increased from 90hz to 120hz.

  the old AMOLED screens are not terribly worse than the new uOLED.

  fov has not improved much, and stereo overlap seems to get even lower (some $1000+ hmds with only 70-80!  glad pimax at least upgraded the super's overlap)

 

so it is frustrating to see all the emphasis put on growing resolution numbers with retina PPD hype...and then the resolution causes a gpu requirement so extreme that there is no way to even play at that new resolution!

 

i like the ideas of pimax better than the trend for small/light hmd but the rendered resolution is too much for me to get the super.  pimax advertised bigger screen panels and i thought this would mean less need for extra rendering to correct the lens focus.  if they can correct these things with the 12k that might make it worth the price.

Edited April 14 by zeppelinzach

[CAG_Sanders]

On 4/13/2025 at 2:44 AM, chiliwili69 said:

Wow, this is really what I feared. The ratio rendered/physical is along the line of previous Pimax headsets:

 

 

You need to render 2.66 times the pixels of the physical panel. So, this is full waste of GPU power spent in this poor geometry of aspheriical lenses.

 

It means that even with a 5090 you only get 45 fps!!

 

 

I had some hope in the device which is the only one giving good FOVs and DP cable.

All the other solutions (BSB2, Meganex 8K ) are MicroOLED based with small FOVs or has not DP cable.

 

I have no problem in getting a 5090, but no, even with that you are at 45fps in IL-2.

 

That´s really bad.

 

Let´s see what default resolutions we will have with the Air Dream and that new pancake lenses (but forget this FOVs)

 

Is this your actual usage or from a review? Still waiting to get hands on mine. 

[chiliwili69]

3 hours ago, zeppelinzach said:

even some fresnel lens HMDs from before 2020 have better clarity than new models.

 

Yes, for example Index has been one of my favourite headsets, just for PCVR, with nice FOVs and DP cable.

At the end of the day, all I want is just an Index 2 with 2.5K per eye resolution to play IL-2.  is this asking too much....????

But I fear that Deckard will be a more complex device

3 hours ago, zeppelinzach said:

so it is frustrating to see all the emphasis put on growing resolution numbers with retina PPD hype...and then the resolution causes a gpu requirement so extreme that there is no way to even play at that new resolution

 

Well, it is true that for IL-2 (where Pimax FFR crashses and DFR is not implemented) the only option is to use QuadViews which is a layer I would want to avoid. 

 

For other games/sims where DFR is implemented and with a 5090, then there are better chances to reach 72 or 90 fps. But I only play IL-2 in my scarce time.

 

But my point is the lenses design and geometry, and those extra pixels you need to render to counteract the lens distortion and to get that PPD. I am really surprised this parameter is not reported in any VR database, like FOVs or panel resolution.

[chiliwili69]

38 minutes ago, CAG_Sanders said:

Is this your actual usage or from a review? Still waiting to get hands on mine.

 

No, I have not the Crystal Super, I got those numbers from the minute 8 of the video above of FlightSimGuy:

 

At least, in a fact like this, I can trust on him. 

 

I previously had the Pimax Crystal Light, and it had also a very high rendered/physical ratio. Exactly 2.65, so it looks like the lenses used for the Super are very similar if not the same.

[Dagwoodyt]

12 minutes ago, chiliwili69 said:

I am really surprised this parameter is not reported in any VR database, like FOVs or panel resolution.

scant profit derived from delivering bad news😉

[CAG_Sanders]

O ya makes sense. Assuming it behaves like the og crystal you don't have to run at that resolution to get great visuals. Also I have found using OpenXR makes a significant improvement on performance compared to steamvr. So at least for flight sim I wouldn't throw the baby out with the bath water just yet. I will report what my performance looks like with settings and my setup to be transparent. 

[chiliwili69]

Interesting to know that the BSB2 has FOVs better than Quest3 according to this review:

 

 

He didn´t mention anything about performance requirements.

 

 

[CAG_Sanders]

Be carefully with his reviews. Very much a master salesman. His chart is misleading in many ways, but one that stands out the most is the fov. If the max numbers people are getting out in the wild are about 110 for quest 3, 108 for BSB2 and 127 for standard mode on the super his chart is not accurate. It calls into question the whole chart and as a heavy user of Q3 and the Original crystal, the centre focus clarity is way better on the crystal than the Q3 I have used. So unless the super is worse than the original, this chart doesn't make sense. Edge to edge is better but not centre focus. Once again I think he is a salesman at heart and loves pancake lenses, and maybe some nice partner money for good reviews. I will say the BSB2 does appear to be exactly what the original BSB should have been and looks like a great product for what it is. 

Edited April 15 by CAG_Sanders
Edit for typo on the BSB2 fov, should be 108 not 118.

[chiliwili69]

1 hour ago, CAG_Sanders said:

It calls into question the whole chart and as a heavy user of Q3 and the Original crystal, the centre focus clarity is way better on the crystal than the Q3 I have used.

 

The wording of "Centre Focus", as explained in the video, refers to the Sweet Spot, which is the area in which your heaset and your eyes are in the correct position to be focused. It doesnot refers to the visual detail you get at the centre.

I know the Q3 has a very good (it means large) sweet-spot, meaining that it doesn´t matter if your headset is positioned a bit up/down or left/right, you still get a focused image.

In contrast, I had the Crystal Light, and I know that the sweet spot is smaller, you need to position the headset in a precise position, otherwise you are not focused.

 

If you refer to details, he mention afterwards that the visual detail is much better in the Crystal Super:

 

 

But I agree with you that all those youtubers has to be taken with a grain (or dozen of grains) of salt since all are subjetive feelings.

 

At the end I resolved to try BSB2 myself, to know my subjetive feelings as well. Which are what it matter for me.  😉 

[dgiatr]

1 hour ago, chiliwili69 said:

 

 

 

Does the above schematic has been made taking into consideration that quest3, Bsb and Pimax Super work on the same resolution or that each one would work on its best possible resolution?

[dgiatr]

1 hour ago, chiliwili69 said:

 

The wording of "Centre Focus", as explained in the video, refers to the Sweet Spot, which is the area in which your heaset and your eyes are in the correct position to be focused. It doesnot refers to the visual detail you get at the centre.

I know the Q3 has a very good (it means large) sweet-spot, meaining that it doesn´t matter if your headset is positioned a bit up/down or left/right, you still get a focused image.

In contrast, I had the Crystal Light, and I know that the sweet spot is smaller, you need to position the headset in a precise position, otherwise you are not focused.

 

If you refer to details, he mention afterwards that the visual detail is much better in the Crystal Super:

 

 

But I agree with you that all those youtubers has to be taken with a grain (or dozen of grains) of salt since all are subjetive feelings.

 

At the end I resolved to try BSB2 myself, to know my subjetive feelings as well. Which are what it matter for me.  😉 

I mean that if you want on purpose to be stuck to about 3000 resolution for better compromisation between spotting and idying then what would be the best headset between the three of them...

Edited April 15 by dgiatr

[chiliwili69]

1 hour ago, dgiatr said:

Does the above schematic has been made taking into consideration that quest3, Bsb and Pimax Super work on the same resolution or that each one would work on its best possible resolution?

I don´t think he has tested each on the same resolution since it will not make sense to raise the Quest3 to a supersampling 836% to reach the deafault resolution of the Crystal Super (78 million pixels).

As he said, it is a subjetive feeling what he reports, not measured by a benchmark or tool. I think he has just used default resolutions for each one.

[chiliwili69]

47 minutes ago, dgiatr said:

I mean that if you want on purpose to be stuck to about 3000 resolution for better compromisation between spotting and idying then what would be the best headset between the three of them...

This is an interesting question, since the rendered resolution is the usual limit imposed by the GPU in high resolution devices.

Setting the bar at 3000x3000 is 18million pixels which is very close to what a Reverb G2 was doing at default resolution (19.5 million pixels). Something on the limits of a 3080 at 90Hz.

For 18 Million pixels :

For the Quest3 at 72Hz you will use 192% supersampling

For the BSB2 at 75Hz you will use 72% supersampling

For the Crystal Super you will use 24% supersampling (so clearly underutilizing the device)

 

For that resolution, for idying I think you will get better plane definition with the Quest3 or perhaps the BSB2. I will tell you that in July. 😉 

I did a similar comparison (but with 16.6 Million pixels) with Quest3 and Crystal Light:

https://forum.il2sturmovik.com/topic/88605-quest3-vs-crystal-light-through-the-lens-pictures/

 

 

 

For Spotting I really don´t know, I was experimenting the same with Crystal Light vs Quest3 rendering high and low resolution:

https://forum.il2sturmovik.com/topic/88605-quest3-vs-crystal-light-through-the-lens-pictures/#findComment-1319486

 

[Dennus]

First post on the forums, so first I want to say hello to all you VR guru's. Your posts are mighty informative.

 

Currently I own a quest 3 and managed to get a new rtx5080 for cheap (well, still 1190 euro, so cheap is relative, as is with all these toys). I was coming from the rtx4070ti super.

 

After much deliberation I pulled the trigger and ordered a Crystal Super last week. We will see what the rtx5080 can do with it. I'm managing my expectations since it's by no means comparable to what a rtx4090 can do let alone a rtx5090.

 

I overclocked it slightly with +300/+1500 which give me a significant increase on 3Dmarks SteelNomad benchmark.

 

I will post my findings and best settings after I did extensive tweaking and testing.

 

Special thanks to Chiliwili69 for his all his testing and sharing insights on VR in IL-2.

[chiliwili69]

In this video, it is adressed exactly that elefant in the room.

He initially run some games, including MSFS and DCS with the full blown resolution of 78million pixels using a 4090.

Later he reduced the pimax multiplier to 0.75. This multplier acts in each axes, so 0.75*0.75 is 0.56, so it renders at 4680x4708 which is 44 million pixels, which gives better frametimes. But still some games needs extra settings to unload GPU.

Since DFR is not possible in IL-2, I assume people will need to play with settings, resolutions and upscaling techniques. It is the price to pay to get the best visuals. 

 

 

 

20 hours ago, Dennus said:

Currently I own a quest 3 and managed to get a new rtx5080 for cheap (well, still 1190 euro, so cheap is relative, as is with all these toys)

 

It is a good price today, for sure you will need all extra power needed for the Crystal Super.

 

Regarding benchmarks, it is better to refers to 3DMArk FireStrike Ultra since it uses DX11 which is what is used by IL-2 engine.

 

You can also run the IL-2 SYN_Vander tests just for CPU and GPU before your Crystal arrives, so it will be the first 5080 reported in that IL-2 list.

[CAG_Sanders]

Finally got my super after being in customs for 2 weeks. Fired up Il2 using OpenXR via opencomposite dll switch. Just like the og crystal I adjusted the resolution, ended up with 3900x3900ish and was able to get 72 fps locked with all settings maxed other than msaa, which I do not use in VR. The image quality is incredible, and soo bright it is mind blowing. The fov is a nice to have but honest the clarity, colors and brightness were the most mind blowing for me. 

 

There is no reason to run the headset with "native" resolution for the distortion profile. The quality between "native" and my adjusted resolution is practically not perceivable. I find long as you use higher than the panel actual specs the headset does a great job adjust for the distortion profile with what you give it. 

 

It does have CA which really only rears it's ugly head in menus so isn't a big deal to me. I don't actively look for mura and it is not noticable on my headset when just playing. Edge to edge clarity is good but not complete on the edges, also I do not try to read things out of the corner of my eye so not a big deal for me. Vast majority of the lens is super clear. 

 

I am running a 4090 and 9800x3d with both over clocked, but I have no doubt with most higher end systems you could get good performance in IL2. Also works great in DCS with DFR. Going to try it out in CLOD soon.  I don't play any other VR games so nothing to add there. 

 

I would say it's a home run for flight simmers. Assuming quality control stays solid. 

[Varibraun]

18 hours ago, CAG_Sanders said:

Finally got my super after being in customs for 2 weeks. Fired up Il2 using OpenXR via opencomposite dll switch.

 

Mine came in last week too, but am just now having a chance to really try it out.  Quick question - are you using XR Toolkit with it? 

 

Mine does not appear to changing resolution in XR Toolkit without making "override" changes in XR Toolkit.  It may be that the Pimax Play resolution is too high to be automatically recognized there.

[CAG_Sanders]

Yes I use the toolkit to override the resolution for Il2. 

[chiliwili69]

19 hours ago, CAG_Sanders said:

Just like the og crystal I adjusted the resolution, ended up with 3900x3900ish and was able to get 72 fps locked with all settings maxed other than msaa

 

So, this is equivalent to put 40% SS in SteamVR, rendering about 30.4 million pixels at 72Hz. So the limit is about 2.2 GPixels/s for a 4090. Good to know.

 

It is also interesting to know that you don´t see extra visual details when you raise the resolution above 3900x3900. 

[CAG_Sanders]

To clarify I I could probably go higher and still be locked at 72. Just was a quick initial testing and was happy with the results. Will do more testing and tweaking when I have more time. 

 

With the visual quality of IL2 it is difficult to see noticeable improvements. In a game that has higher visual fidelity it might be more obvious. 

[Varibraun]

My very early initial impressions track with what @CAG_Sanders has reported above.  I am running a similar system (4090 GPU, 13900 CPU).  A few other preliminary observations:

 

-IL2 FOV (the number that reflects with "backspace" that shows up beside the FPS counter) reports 117 vs. 94 with my OG Crystal.  If I remember correctly, the IL2 FOV reported about 10 degrees lower than actual FOV for my OG Crystal, so the IL2 117 is probably actually the 127 the YouTube reviewers have measured.

 

-My OG Crystal would run IL2 extreme graphics settings with FXAA at 90Hz consistently at 4200ish setting.  Without changing anything and allowing the Super to run Pimax 1.0 "native" which I think is approx 6000x6000 my FPS dropped to 35-56 running in 90Hz mode.  Unsurprising, and a no-go, so have been experimenting with XR Tool Kit. 

 

-In a 2 v 2 Quick Mission on the Rhineland map, I have been able to maintain the 72Hz setting at 5000x5000 and 4600x4600 in 90Hz without a problem (I haven't tried 90Hz yet  ).  This is with IL2 extreme graphics settings and 2xFXAA.  However, as @CAG_Sanders noted above and @chiliwili69 previously predicted, for IL2 at 5000 there may be a slight improvement in graphics over my OG Crystal, but nothing "earth moving" for me so far on the graphics as compared to going from the G2 to the Crystal.

 

-I will continue to experiment with settings over the next week and check out a couple of other games.  Right now, I am thinking I will probably keep the headset based on a combination of the following factors:

*FOV improvement (the extra 23 degrees is noticeable over the OG Crystal).

*No battery is a big improvement for me.  

*Weight is better, but I didn't find the OG Crystal uncomfortable after modding with a Pimax headstrap and the StudioForm foam (I will order this once it is available for the Super).  Oddly, I don't like the optional headstrap with the Super and so far prefer it with the the basic one.

*DCS and future games that will support features allowing the headset to increase resolution without FPS loss (i.e. locking in this as my headset for the next few years).

*OG Crystal owner's discount.

*Tariff uncertainty in the US (preorder saved that potential issue for this headset).

*While some Mura and CA are present, for me personally they are not factors that I see without looking for them.

 

I am still early in the 14 day return window, so I have some time to keep testing my impressions in a non-technical way (I am not a "benchmark" guy, which is why I appreciate everyone here who provides that info for me to consider). 

 

 

 

Edited April 29 by Varibraun

[Broadway]

Thank you to the posters on here , I’m interested to continue to hear the feedback , i too have the OG Crystal with the Super due in June……. . I’m running 7800x3d (possible may upgrade) and a 4090 (no immediate plans to upgrade due to availability and cost/ power)

I like the idea of the increased FOV , and no battery, and am keen to see how much res i van get away with to run at 72, also if there are any meaningful improvements in the quality (brightness / contrast / colours) vs other HMDs 

Thanks all 

 

[chiliwili69]

On 4/28/2025 at 3:12 AM, Varibraun said:

I have been able to maintain the 72Hz setting at 5000x5000 and 4600x4600 in 90Hz without a problem (I haven't tried 90Hz yet  ).  This is with IL2 extreme graphics settings and 2xFXAA

 

For the 72Hz this is about 3.6 Gpixels/s and for the 90Hz is about 3.8 Gpixels/s.  This is way beyond what I thought a 4090 can deliver! (are you sure you have a 4090 and not a 5090?  😉 ). Are you using any upscaling techniques or any Pimax FFR?

 

@CAG_Sanders was obtaining initially 3900x3900 at 72Hz , this is 2.2 Gpixels/s but he also said there could be room for more.

 

For performance monitoring either the overlay fpsVR or the overlay of MSIAfterburner are good to know the GPU load, fps, framtimes, while you are playing.

[CAG_Sanders]

Keep in mind we don't all use the same testing formula. Just so aware, I run my test on Normandy with my flight of 4 109s, and 3 other flights of 4x B-25s at 2000 meters.

 

I have found this to be a great stress test for the engine. If I used less planes and a better performing map I could most certainly use a higher resolution, but like I said perceivable difference on higher than 3900 is very limited. The fidelity of the visuals on GB is just not there.

[chiliwili69]

25 minutes ago, CAG_Sanders said:

Keep in mind we don't all use the same testing formula. Just so aware, I run my test on Normandy with my flight of 4 109s, and 3 other flights of 4x B-25s at 2000 meters.

 

I have found this to be a great stress test for the engine. If I used less planes and a better performing map I could most certainly use a higher resolution, but like I said perceivable difference on higher than 3900 is very limited. The fidelity of the visuals on GB is just not there.

 

Yes, the map and density of objects could influence on the GPU load since there will be more polygons to treat and render, but what I was a bit surprised is the perceivable visual quality above 3900.

I believe the render resolution can not dynamically changed while you are playing or in pause mode, so it is hard to keep memory of you see at one resolution and then restart the game with another resolution.

 

What I have been doing in the past to compare the visual quality of one VR device at diferent resolution or also two diferent VR devices was to create my own set of short recorded tracks (one for IDs, one for spotting, one for visual detail, etc) and then using a camera. I also compared with a monitor 4K and it is not a problem of the GB visuals (which are very detailed):

 

 

 

 

For example, @DBCOOPER011 was shoing the gain in the visual detail with the original Crystal:

 

3 screenshots of the crystal at different resolution. 2888x3100, 4312x4628 and 5604x6016.

 

I would expect to have the same gain with the Crystal Super, just in a paused moment, without taking into account the performance (it is just a question of buying a 5090)

Edited April 29 by chiliwili69

[CAG_Sanders]

I think we both have a fundamental different understanding or viewpoint of how VR resolution works specifically in regards to barrel distortion.

 

Over sampling a lower resolution screen doesn't not have the same visual quality as a higher resolution screen. Just like over sampling a 1080p monitor to 4k does not look as good as 4k native or even 4k running a lower resolution then super sampling it. The distortion profile is super sampling the panel resolution to account for barrel distortion. 

 

The headset is able to handle the lower resolution and adjust for barrel distortion, partial due to still driving the headset at or above it native panel resolution. Primarily the downside is some softness, which like I said is barely perceivable in game when compared to the full distortion profile. 

 

There is a good leap above the og crystal in my opinion. You just don't need to run it at the full distortion profile. 

 

I will also agree to disagree that GB is highly detailed considering the competition that exists in the space.

 

I do appreciate your insights, and I think you clearly find joy in the minutiae of VR headsets, but even comparing the data from you quest to crystal there is some flaws. For instance the crystal picture of the spot wing, there appears to be some loss of detail in some areas and some increase in other when comparing the 16 to 44. I guess what I am trying to say is it's not a perfect science how VR is perceived. 

[Varibraun]

7 hours ago, chiliwili69 said:

For the 72Hz this is about 3.6 Gpixels/s and for the 90Hz is about 3.8 Gpixels/s.  This is way beyond what I thought a 4090 can deliver! (are you sure you have a 4090 and not a 5090?  😉 ). Are you using any upscaling techniques or any Pimax FFR?

 

@CAG_Sanders was obtaining initially 3900x3900 at 72Hz , this is 2.2 Gpixels/s but he also said there could be room for more.

 

For performance monitoring either the overlay fpsVR or the overlay of MSIAfterburner are good to know the GPU load, fps, framtimes, while you are playing.

 

Yeah, sorry Chili, I knew my "@dburne methodology" would drive you crazy (I do miss him here).  But I did provide you fair warning about my lack of tech/benchmarking expertise.  Honestly, I am probably the last guy that others should rely on for #s analysis.

 

Anyway, those are the resolutions set in XR Toolkit override that get me smooth play without ghosting and hold the IL2 FPS counter mostly steady at 72/90.  @CAG_Sanders is definitely stress testing harder areas/maps/#AC than I am right now, but I did run a couple FC Career missions last night with some heavy flack and 10+ AC dogfights and the lowest # I saw was 85 when at 90Hz and 4600x4600.

 

But for you (seriously, thank you for all of your analysis for over a decade), I will try to see if I can get FPSVR fired up to see if it can provide any insight into why I am seeing that on a 4090 (however, I will tell my wife that you have approved a 5090 purchase for the household in the interest of scientific analysis).  Bottom line right now is that I am not aware of doing anything other than XR ToolKit override, but maybe I set something else and those XR ToolKit #s are not an accurate report of the actual resolution?

 

6 hours ago, CAG_Sanders said:

There is a good leap above the og crystal in my opinion. You just don't need to run it at the full distortion profile. 

 

One area where I think I may be seeing another improvement is better detail at a distance.  Is this your experience, I am always a little concerned about placebo effect?

[CAG_Sanders]

@Varibraun I will take a look at distance detail and see what I think. My gut is yes it is better but I have spent most of my time looking in the intermediate and medium range.

[dgiatr]

1 hour ago, Varibraun said:

(however, I will tell my wife that you have approved a 5090 purchase for the household in the interest of scientific analysis). 

🤣🤣👍

[TCW_Brzi_Joe]

On 4/8/2025 at 9:47 PM, TWHYata_PL said:

so the "lab mode is something in PimaxPlay or just another HW revision?

"Lab mode" means it is experimental (beta) feature. You have to turn on betas in pimax play, and install latest pp.

 

Please somebody with pimax super try to spot one pixel outside of 10km in multiplayer (fly on deck to known busy target and look in that direction, the best angled trough side window). I am afraid in this case lower resolution headsets beat such a high resolution beasts... 

Edited April 30 by TCW_Brzi_Joe

[Varibraun]

On 4/29/2025 at 12:40 PM, chiliwili69 said:

For performance monitoring either the overlay fpsVR or the overlay of MSIAfterburner are good to know the GPU load, fps, framtimes, while you are playing.

 

Hi @chiliwili69, as promised: Below are fpsVR stats with SteamVR, Pimax & IL2 settings from a Steam VR session last night (fpsVR didn't want to play in Pimax OpenXR, so I switched to Steam VR @ approximately the same 4600x4600 resolution).  It will also probably give you some more familiar stats since I think you fly mostly in Steam VR.  These are in a 2v3 Quick Start Mission on Rhineland Map, but I also flew over some towns and did some ground attack in the 15 minutes.  Also, the session included some paused and loading time that seems to bring the average framerate down just a bit - I never saw anything less than 90 while flying.  I am sure these probably explain better to you what I am seeing, but I don't think I am currently doing any upscaling.

 

Spoiler

 

 

 

 

 

 

 

 

 

 

 

 

Edited April 30 by Varibraun

[chiliwili69]

On 4/29/2025 at 7:48 PM, CAG_Sanders said:

Over sampling a lower resolution screen doesn't not have the same visual quality as a higher resolution screen

Yes, We agree on that.

On 4/29/2025 at 7:48 PM, CAG_Sanders said:

Just like over sampling a 1080p monitor to 4k does not look as good as 4k native

You can compare a in a monitor 1080p with no supersampling (so you render 1920x1080 pixels=2million pixels) and then the same scene with 200% supersampling (4million pixels rendered but displayed in a 2 million pixels monitor, ie a 1080p monitor). You will notice that the supersampled image will have a bit better quality.

If you increase the SS to 300% it will be a bit better (but less gain quality than before) an so on, until you will not notice any diference by increasing supersampling.

You can for example do a 400% SS, what is equivalent to to the render at 4K but display it at 1080p. These are the techinques that NVIDIA (DSR) and AMD (VSR) were giving for flat 1080p monitors. (but can also applied to 4K monitor and render at 8K)

You get better visuals but a higher GPU load.

This is an example of the imoprovement:

 

Of course, rendering at 4K and using a 4K monitor will be much much better.

Edited May 2 by chiliwili69

[chiliwili69]

On 4/29/2025 at 7:48 PM, CAG_Sanders said:

The distortion profile is super sampling the panel resolution to account for barrel distortion.

Yes, in VR all the headsets (except some headset with pancake lenses like Quest3) has to render at a higher resolution to compensate the distortion introduced by the lenses.

The software distortion applied to the image before is sent to the panel is called barrel distorsion, and depending on the geometry, FOVs and type of lenses this distorsion is more or less severe. Because the barrel distorsion expand the pixels in the center of the image, it will contain less rendered pixels in the center and that´s why almost all devices has a default resolution higher than the physical panel.

Here it is explained better:

 

[chiliwili69]

On 4/29/2025 at 7:48 PM, CAG_Sanders said:

The headset is able to handle the lower resolution and adjust for barrel distortion, partial due to still driving the headset at or above it native panel resolution. Primarily the downside is some softness, which like I said is barely perceivable in game when compared to the full distortion profile.

 

At the end of the day, for any given VR device, the less pixels you render the less detail you get.

 

You can take the Crystal and render at 50%, or at 100% or at 150%. And this is what @DBCOOPER011 did with the image above.

 

Every manufacturer set the default resolution at a value they consider acceptable to get the required PPD at the center of the image. But, of course, the user can run below or above, depending on the GPU power.

 

I was experimenting with this in the past with the Rift: https://forum.il2sturmovik.com/topic/30771-how-much-ss-is-desirable-image-quality-samples/

 

An also with the Reverb G1: (BTW, the G1 had the default resolution really low at the begining, later they corrected it to have 9.5 miilion pixels at 100%):

https://forum.il2sturmovik.com/topic/53689-hp-reverb-supersampling-100vs188vs300-through-the-lens-pictures/

 

 

On 4/29/2025 at 7:48 PM, CAG_Sanders said:

There is a good leap above the og crystal in my opinion. You just don't need to run it at the full distortion profile

 

Yes, of course, you have a panel with almot twice the number of physical pixels (OG Crystal 16.6 Million, vs. Crystal Super 29.5) and with exactly the same render/physical ratio which is 2.66.

 

So, the visual is going to be always better in the Super, for any given amount of rendered pixels.

The difference would be that above 50 million rendered pixels (or a number around) the OG Crystal will not give you more details, whereas the Crystal Super will still give you more details. The only problem is to have a GPU able to render more than 50 million pixels.

[chiliwili69]

On 4/29/2025 at 7:48 PM, CAG_Sanders said:

I will also agree to disagree that GB is highly detailed considering the competition that exists in the space

The textures used in IL-2 are very good, you only need to lean your head and get closer to the gauges in the cabin, the level of detail is very high. The only problem is to get a device able to reproduce them. So, IL-2 is not lacking detail, it is just the VR devices not giving enough PPD.

[chiliwili69]

On 4/29/2025 at 7:48 PM, CAG_Sanders said:

I do appreciate your insights, and I think you clearly find joy in the minutiae of VR headsets, but even comparing the data from you quest to crystal there is some flaws. For instance the crystal picture of the spot wing, there appears to be some loss of detail in some areas and some increase in other when comparing the 16 to 44. I guess what I am trying to say is it's not a perfect science how VR is perceived.

 

Yes, it is not easy to get through the lenses pictures, normally I shoot about 12 pictures per case and select the more focused one, but I just do that to avoid my own placebo efect (or bad visual memory when comparing two devices A-B). I don´t want to be cheated by my own senses and feeling when a camera will tell me with more precision what image is more detailed.

 

Regarding the image you mention, the important thing is to compare the focused area (normally I try to focus the center and provide the crops of the focused area).

I don´t know why But I forgot to do it for the image you mention. I have uploaded those three crops in the corresponding thread:

This is the Crystal Light:

16Mpx: This is PimaxPlay Balanced resolution (it is 0.75) with SteamVR at 66%SS, this is 2644x3128 per eye, so 16.5 Million pixels. (or Pimaxplay 1.0 and StreamVR 38%)

44Mpx: This is PimaxPlay maximun resolution (it is 1.0) with SteamVR at 100%SS, this is 4332x5124 per eye, so 44.4 Million pixels. 

 

What I mean with these Crystal Light example is that if you do the same with the Crystal Super (running it at 50% and at 100%)I would expect you will get a noticeable increase in the details and image quality.