ackattacker

I read somewhere that to fully reproduce human vision in a VR headset would require 500 megapixels. But that if you perfectly implement foveated rendering, only about 7 megapixels need to be actually rendered.

Also, small sweet spot and outside the sweet spot causes geometric distortions/ pupil swim. People call this guy a Pimax hater but I think he’s just being realistic. A $1600 device that’s a year late and still feels like a beta product. Which, honestly, is typical Pimax. He’s upfront that he’s not a simmer, but really the headset is sold as a do-everything, what with the wireless and standalone capabilities. It really seems sub-optimal for those uses even after the software side is sorted. I’d much rather watch his reviews than Sebastian’s “everything is wonderful” gush-fests.

Yeah I could see something like virtual desktop allowing PC streaming. But the resolution would be challenging over wifi for a good experience. And early descriptions from folks who got to try it, the FOV is not particularly large. By the time this releases next year, we may see other headsets with similar resolution better optimized for PCVR with lower price tags. I don't ever see this as being a desirable headset for DCS. That said, I could be proven wrong. If enough talented developers tackle the problem, perhaps some sort of foveated transport would allow the experience over wifi to be actually good. I'm not holding my breath though.

There is no way this $3500 device is going to take VR mainstream. Huge disappointment. Even the resolution is lower than reported, price is higher, 2 hour battery life, and it basically projects 2D iOS apps onto your field of view. Which is better than a monitor because…. As expected no mention of PCVR compatibility. Which means a solid nope.

4000x4000 x 2 is about the same as an "8k" monitor... 7680x4320... a 4090 can drive it l, but not AAA games at 120fps... a lot depends on what you are trying to display. Productivity, AR, simple polygon games would be no problem with 2xM1. But DCS would be nigh impossible I think at that resolution. I guess we'll find out!

The 4k x 4k displays in the Apple (made by Sony) are coming soon to other headsets. The big problem right now would be driving that resolution, it's like running four 4k monitors at once. Good, effecient foveated rendering and foveated transport are going to be critical to make that work.

The price is sky high not solely because of the displays, but because of the massive processing power built in. It is intended to run standalone apps, and augmented reality. The display technology is not exclusive to Apple and will make it’s way to other, less crazy priced headsets. That said, it is not totally inconceivable the “Apple Reality Pro” could work with PC games. Just highly unlikely…

The Crystal has about the same number of total pixels as an 8kx. With a 4080/4090 and the MT build it is very possible to achieve a steady 90fps with the 8kx and optimized settings. 60fps is easy even with most everything cranked up.

I have been running very similar settings with my 4090 with good results. It is stable and smooth at 90fps.

It seems obvious to me that the game engine should simply calculate whether an object would be visible to a person with 20/20 vision, and, if not, NOT DISPLAY IT AT ALL at any resolution. If the object calculates as visible, and If a person is playing at a ridiculous low resolution to make single pixel dots much more obvious, the game engine could decrease the contrast of that pixel so that spotting it is of comparable difficulty to spotting a smaller, higher contrast pixel. This is basically what is happening in VR… because in VR the game engine renders a higher resolution than the panel resolution, which is downsampled in the distortion correction. So a single black pixel becomes a single gray pixel in the VR pipeline. So the game engine could apply a similar “downsampling” to 2D, low res players, and thereby make spotting more realistic.

Let me reiterate that I’ve been doing this professionally for decades and that looking for traffic is something I do daily. Nevertheless as a result of this discussion I decided to test this today on my way back from Madrid. Crossing the North Atlantic at FL340 (34,000 feet), there was a fair amount of opposite direction traffic on the same route a few thousand above and below me. TCAS (Traffic Alert and Collision Avoidance System) would pick up the traffic at a distance of approximately 25-30nm, and inform me precisely where to look. At this altitude they were producing contrails. I could easily see the contrails at this distance but could not see the aircraft themselves until approximately 15nm despite the contrails making their location obvious. Later we climbed to FL380 and ultimately FL400 over the Canadian maritimes. At this point traffic was not producing contrails. There where two large wide body aircraft shown on TCAS about 25 miles to the right, one at my altitude and one 2,000’ below. The sun was shining directly on these high-aspect, 260 foot long, 50 foot tall brilliant white aircraft silhouetted against a deep blue sky, and TCAS was directing me precisely where to look. After a minute or so of searching, I was able to see those aircraft, just barely. I found however, that despite the relatively small angular difference between them, if I looked at one aircraft I couldn’t see the other. I had to look in between them to see both. If I took my eyes of them, they where difficult to re-acquire. If TCAS hadn’t been directing me where to look, I would never in a million years have seen them from simply scanning the sky. Low-aspect traffic coming directly at me from the other direction could not be spotted until about 15 miles, again even though I knew exactly where to look. Once they where within 7-8nm, spotting was much easier and I simply had to look in that general direction of sky. Let me reiterate that these aircraft are about as long as a football field and painted brilliant white. And, at least according to my AME, I do have 20/20 vision. Now, imagine instead these aircraft where only 62’ long (the length of an F-14 Tomcat) and painted matte gray. Or, for another thought experiment, imagine standing on top of the Empire State Building (1250’/380m tall) and spotting a die-cast model of an F-14 Tomcat 4.7 inches in length sitting on the sidewalk. That is is being described when spotting an aircraft at 60km. Let me reiterate, *comical*.

Sounds like DCS needs to nerf the guys using 2d monitors then. 60km spotting distance is comical. I found this on Hoggit: Several investigations have been made to determine aircraft target acquisition capabilities. A total of 759 training engagements at the Naval Air Station Oceana Tactical Air Combat Training System (TACTS) range revealed that in 624 of the engagements the pilots first sighted the target as a dot against the background at an average distance of 5.67 nmi (Hamilton & Monaco, 1986; Monaco & Hamilton, 1985). In the remaining 135 engagements exhaust smoke, contrails and sun glint off the aircraft allowed the pilots to detect the aircraft at even greater distances. In the 122 engagements where exhaust smoke was the primary cue, detection distances averaged 7.64 nmi. Environmental and local conditions as well as target type and paint scheme play a significant role in detection distances though. Variables such as background sky or ground coloring vs. aircraft coloring, brightness and directness of sunlight as well as target location vs. the sun and several other variables can either enhance or decrease detection distances. Furthermore, although Hamilton & Monaco found several instances where exhaust smoke was the primary cue, this condition is arguably becoming of decreasing value as aircraft emissions have become less visible over the last decade or two. Table 1 provides a large list of factors that have been shown to affect target detectability. These items were taken from the field evaluations cited in this section as well as Bloomfield & Smith (1982), Boff & Lincoln (1998), Buffett (1986), Costanza, Stacey, & Snyder (1980), and Hoffman (1976). In 1983, Kress & Brictson studied 87 air-to-air engagements at the Yuma TACTS range. Average unaided detection distances for the target F-5 and F-4 aircraft were 3.1 nmi. When the pilots were aided with a head-up display (HUD) symbol that cued the pilot to the target’s location, the mean detection distance grew to 6.8 nmi. Another study that investigated detection distances was Temme & Still (1991). They measured air-to-air target detection distances at the Naval Air Station Oceana TACTS range to see if there was a performance difference between those pilots who wore corrective eyeglasses and those who did not. Those with eyeglasses did not detect the targets until they were about 10% closer than those with unaided vision. Two very closely matched groups of eyeglass and non-eyeglass wearers had average detection ranges of 4.52 and 5.64 nmi respectively when using all detection means including aircraft sighting, target glint, contrails and exhaust smoke. When limiting subjects to aircraft-only detections, the corresponding distances were 4.35 and 5.54 nmi respectively. Although the distinction of glasses vs. no glasses is not of interest to this investigation, it does provide two more data points for detection distance ability. Another study by Hutchins in 1978 at the Air Combat Maneuvering Range (ACMR), which is the earlier name of the TACTS, involved 45 air combat training engagements. The mean detection distance of the A-4 targets was 3.09, with a range of 0.38 to 6.23 nmi. Other studies were done using observers on the ground. With visibility conditions spanning 7 to 10 miles over an 8-day testing period, O’Neal & Miller (1998) found detection distances for approaching T-38 aircraft to ranged from 4.77 to 6.73 nmi. Another ground observer study used 400 visual detections of a T-38 aircraft (Provines, Rahe, Block, Pena, & Tredici, 1983). The aircraft was approaching from a known direction and a distance of 9 miles and mean detection distance was 4.55 miles over the 400 trials. References Hamilton, P. V., Monaco, W. A. (1986). Improving air-to-air target detection. Wings of Gold, 46-48. Kress, G., Brictson, C. A. (1983). Operational Analysis of Visual Skills for Air Combat Maneuvering (NAVTRAEQUIPCEN 80-D-0011/041-3). Orlando, FL: Code N712, Naval Training Equipment Center. Hutchins, Jr., C. W. (1978). The Relationship between Air Combat Maneuvering Range (ACMR) Output Measures and Initial Visual Acquisition Performance (NAMRLSR-79-1, AD-A062 134). Naval Air Station, Pensacola, FL: Naval Aerospace Medical Research Laboratory. Provines, W. F., Rahe, A. J., Block, M. G., Pena, T., Tredici, T. J. (1983). Yellow Ophthalmic Filters in the Visual Acquisition of Aircraft (USAFSAM-TR-83-46, AD-A138 536). Brooks AFB, TX: Aerospace Medical Division, USAF School of Aerospace Medicine

I’m a professional pilot with well over 10,000 flight hours and I can tell you that IRL it’s damn hard to spot a wide body airliner at 16km. If people are using mods to spot at 60km in DCS then they are basically cheating as far as I’m concerned.

OpenXR toolkit will also show you cpu and gpu frametimes, and has logging.

I did run out and get a 4090... now I can achieve 90FPS consistently at native resolution normal FOV (I am using still using FFR), high textures, medium shadows, no MSAA. Even busy multiplayer no problem. It looks and feels great, finally I feel the 8kx is living up to it's promises.

I think the Bigscreen Beyond looks not ideal for sims but a pretty good option. FOV appears to be about 90 both horizontal and vertical which is about the same as the Quest 2 and G2 which are certainly used by many simmers. Biggest issue would be can DCS drive that resolution without eye tracking and DFR, and also peak brightness seems low. Edit: for comparison, nextgenVR measured the Varjo Aero and Quest 2 and got 90/70 and 88/98 hfov/vfov. Certainly some people are happy with the Aero and the Bigscreen Beyond seems to offer much better VFOV, similar HFOV, and similar resolution/ppd.

I tried 60Hz mode without upscaling and couldn't take it. 75Hz is minimum for me for an enjoyable experience, 90Hz is noticeably better. I will drop all other visual settings to achieve at least 75Hz. Everyone is different. I also cannot stand motion reprojection even when it's working properly. The double images drive me batty. The MT build is a literal game changer for me because previously I would be CPU limited trying to achieve steady 75fps on busy missions. Now that the CPU is not the limit I may have to sneak out and buy a 4090. This is an expensive hobby.

I haven't had a lot of time to optimize it but I have OpenXR/PimaxXR/OpenXRTK working marvelously on my system with multithreading. Only thing is motion reproduction is a mess, but isn't needed with the proper settings. SteamVR works better with motion reproduction but is a much worse experience overall. 5800x3d/3080Ti/32GB/Pimax 8KX DMAS (2076) Pitool V1.0.1.284 Headset firmware V2.1.255.2003 (this is the 120Hz beta firmware) PimaxXR v0.3.2 OpenXR Toolkit v1.3.0 Pitool settings 75Hz, Normal FOV, Render Quality 1, FFR OFF, PP OFF, Smart Smoothing OFF, vertical sync OFF, Hidden Area Mask ON DCS Settings VR Preset, PD 1.0 OpenXR toolkit settings Target Frame Rate 75, Upscaling NIS 80% sharpness 20% FFR Performance/Wide Turbo ON Launch DCS-MT with --force_enable_VR --force_OpenXR Initial testing the system maintains 75fps rock steady with about 10% GPU headroom. This is on single player missions, I have not yet tested multiplayer. The overall experience is smooth. I have no issues with jitters/microstutters/menus etc. It all works fine. Like I said this is not highly optimized settings, just settings that work as a starting point. If you have a 4090, I suspect 90fps is easy peasey probably without upscaling.

Dlss works with msfs in vr no reason it couldn't work in DCS with proper coding.

Those numbers are compelling if you are a MSFS player. Unfortunately DCS is it’s own animal and not nearly as optimized for multi core CPU’s. Note how the MSFS numbers scale dramatically with core count. This doesn’t happen with DCS. DCS is all about keeping that one main core fed, and the cache helps with that dramatically.

With no budget constraints option 3 is the most sensible IMHO. Most future proof platform. If the 7800x3d is anything like the 5800x3d it will blow Intel away in sim games like DCS. Right now though AM5 motherboards are a huge price premium. Might make sense to just hold off the upgrade all together until early 2023, when cheaper AM5 motherboards will be available along with the 7xxxx3d.

What I am saying is that the x3d doesn't exhibit this behavior. The max boost is hardware locked and cooling the cpu "extra" behind what is required to avoid throttling is just a waste of money to no benefit. This is even more true with DCS which is basically a single core workload that doesn't push the thermal limits at all.

I would just run any decent air cooler in the x3d. You can't really overclock it so an AIO will give zero performance enhancement. Personally I run water cooling on my GPU, top exhausted, which keeps the heat out of the case. My x3d under gaming loads stays nice and cool with a simple tower cooler and a single lazy fan pushing the heat out the back. Extremely quiet and case temp is basically room temp. Cpu temp is maybe 60-70c.

My suspicion is that if you are a VR user, especially Pimax 8kx, the 4090 will provide noticeable gains because it's just so many pixels that the GPU can really matter. Problem is normally this game is very much CPU bound, especially multiplayer. I was more excited about the new Ryzen series CPUs. Game benchmarks are trickling in and unfortunately doesn't look like a major uplift compared to 5800x3D. Looks like we'll have to wait for the rumored 7800x3D. I'm kind of thinking next year 7800x3D and 4090Ti will be the performance king, and it's worth holding off on any upgrades because right now for DCS I think the gains will be minimal.

Here is a cut an paste from a moderator on another forum which maybe this guy knows what he's talking about ("Kojack" on OculusVR forum from Oct '21). Since DCS supports OpenXR, and SteamVR supports OpenXR, and Pimax supports SteamVR, the transitive property says that Pimax supports OpenXR ?? Except really I believe what is happening is that the *OpenXR toolkit* is interfacing with SteamVR, which is compatible with OpenXR plugins. SteamVR is then wrapping the Pimax Unity SDK. So really you are not using the OpenXR runtime, you are just using the OpenXR interface of SteamVR. I don't see a huge advantage of doing this versus, say using the VRPerf toolkit. There are three main VR apis for PC headsets: OpenVR (SteamVR is an implementation of it), Oculus (often called OVR) and WMR (for headsets like the HP Reverb G2). SteamVR includes plugins that redirect OpenVR games to Oculus or WMR (SteamVR itself has no built in support for direct access to Oculus or WMR devices, it uses the native SDKs). OpenXR acts as a common interface that bypasses the native SDKs. Oculus, Valve and Microsoft provide OpenXR runtimes. There can only be one runtime registered in the system at a time, that's what an OpenXR game will use. If you have the Oculus OpenXR runtime registered, all OpenXR games will be handled by the Oculus drivers, (same for the other two). If you have two kinds of headset installed, such as a Quest 2 (with Link) and a Reverb G2, registering the Oculus runtime means all OpenXR games go to the Quest 2. Registering the WMR runtime means all OpenXR games go to the Reverb. In the case of SteamVR, it's slightly different. Registering SteamVR/OpenVR's OpenXR means all OpenXR games go to SteamVR, which then either handles it directly (for headsets like Index or Vive) or redirects to the Oculus SDK or WMR SDK. Of course none of this actually happens if the game isn't built for OpenXR. The only one I definitely know of is Microsoft Flight Simulator. So 99.999% of current VR games will be unaffected by which OpenXR runtime you have.