I really enjoyed large FOV in Boneworks. Hadn’t tried it for a long while. Found the large FOV to really really increase immersion. That’s near as damn my real world FOV. I manually changed the Steam setting before just to be sure on the res. Takes some graphics grunt though!
Think fixed foveated rendering would be ideal for that. No need to render all the periphery pixels. I only have a 1080 ti though. One for the next generation!
Oooh, that’s nice to hear!
I must have missed that patch note.
Depends on which density you mean. :7
At least last I looked, the bitmap size piserver requests on SteamVR at PiTool quality=1.0 for “large” FOV is not as much larger than for “normal”, as it would need to be, for the “normal” part of the “large” area to remain as dense. :7
The “default” render resolutions are more or less arbitrarily chosen, and presumably picked for performance more than qualitative consistency. :7
I couldn’t find the setting in SteamVR in Developer Web Console or Developer Settings. Where is the MaxRes setting?
http://127.0.0.1:8998/console/index.html just gives a “can’t reach this page” error.
Apparently Valve changed the URL, you probably need to run it from
SteamVR->Main Menu->Developer->Web Console
Then use settings command as mentioned above. For example, typing
settings max
will show
Fri Dec 20 2019 01:00:20.106 - [Console] Settings:
Fri Dec 20 2019 01:00:20.106 - [Console] Usage: settings [name] [value]
Fri Dec 20 2019 01:00:20.106 - [Console] ----------------------------------------
Fri Dec 20 2019 01:00:20.106 - [Console] steamvr.maxRecommendedResolution: 8192
Fri Dec 20 2019 01:00:20.106 - [Console] steamvr.mirrorViewGeometryMaximized: true
Fri Dec 20 2019 01:00:20.106 - [Console] ----------------------------------------
Fri Dec 20 2019 01:00:20.106 - [Console] 2/239 Matches.
on my machine.
Oh my. Well, so that’s the culprit then now ? Any idea what to do? Prolly nothing we can fix with some json magic? :>
Are you talking about Native or PP mode? In PP mode it seems to me that the Large res is proportionally wider and has the same height as, for example, the Normal FOV.
Normal PP res (PiTool RQ=1, SS=100%) is [3852, 3288]
Large PP res (ditto) is [6948, 3288]
What might be the detriment to the image quality however is the reprojection from the parallel plane to the native (panel) plane.
Helpful thread. I just today encountered this particular bug when trying to work out a skyrim vr parallax issue (some shadows and water being just every so slightly outside the correct focal plane). So I tried turning on the “Parallel Projection” option, which Skyrim VR isn’t supposed to need but which I read somewhere might fix those minor issues (it does). Turns out this is the same bug that causes the problem with PCars2 (which does require Parallel Projection be turned on). So this issue isn’t specific to PCars2 as I had thought, but rather is an incompatibility between two specific PiTool settings.
The low down is this: If you are using Large FOV with Parallel Projection ON, then the SS settings will be ignored and the game will run at an extremely poor resolution. Outside of that one scenario everything seems to work fine. So if you have to use Parallel Projection with a particular title don’t also bother trying to use the Large FOV setting for that game.
That might not be true of all games. I tried Large FOV in Elite D last night and could barely tell the difference between Large and Normal FOV. I switched it dynamically in PiTool while ED was running. I’ll try it again tonight, starting in Large FOV.
Really? I don’t recall seeing it anywhere in the patch notes
Well, true, my sample size is small, but it is accurate across what I’ve got to try. Someone should try it with a larger set of games if they have a chance and see if it holds across other titles. If this does indeed turn out to be a general bug that’s probably good news (easier to fix a single fundamental bug than something effecting multiple titles across a variety of game engines. Or at least so I would think.)
So I tried it again and made sure to restart SteamVR. PP needs to be on for ED. The in-game visuals were essentially the same, other than the wider FOV. The framerate was noticeably reduced, so I’ll continue to play using Normal FOV.
Here are the resolutions I play at, PiTool quality is set to 1.0, and the vertical res may vary from others, since I have a fairly large vertical offset in PiTool (+3.0 and +8.0).
As you can see, Large FOV has noticeably higher X resolution.
Well… my SS settings are ignored even without PP, always in large FOV
Hmm, supersampling should do it - just a matter of increasing the render target bitmap size, after all.
If SteamVR supersampling does not work with a particular title, then quality in PiTool, which determines what SteamVR’s 100% is, should, and if that doesn’t work, one may have to see if there’s any setting of the game’s, that can do it - if not in exposed user settings - then perhaps in its “.ini” or equivalent files… Unreal Engine 4 titles, in particular, are infamous for not giving a damn about one’s SteamVR superampling setting (but PiTool quality does seem to take – don’t know where UE4 gets that size, if not from the usual SteamVR function call :7).
That really sounds like a case of exceeding the SteamVR “maxRecommendedResolution:”, that Neil mentioned , and being capped at that maximum size; PP swells the viewplane significantly, in both axes, as Risa explained.
Of note here, is that the SteamVR settings window has for some reason been showing uncapped numbers under the supersampling sliders, even though the actual recommended render target asked of the game has been capped.
However: If Valve have indeed now removed that cap, as JoCool intimates, this shouldn’t matter. (there was a different thing fixed in a recent SteamVR beta update, that fixed the numbers not updating correctly in some cases- it wasn’t by any chance that you saw, Jo?)
(There have also been talk about Pitool raising the limit behind the scenes, when one launch SteamVR from Pitools, but I have never noticed that effect myself. Wonder how high, then - would be unfortunate if it overwrote one’s own, even higher, custom value… :7)
If is quite possible that some softness will also be introduced by the extra factor to the transformation, with PP – not so much if it is all done in a single step - just modifying the predistortion shader for the lens, to take the offset viewplanes into consideration; More so if it is for some reason done as a separate operation.
Things like cubemap reflections and shadowmaps are troublesome in VR to begin with, especially on surfaces with normal maps (Help me Obi Ray Tracing, you are my only hope. :P), but additionally I have come across quite a few cases where they turn out particularly off, even with parallel projections on - both with a Pimax headset, and the also canted Valve Index. I am not sure what happens, but something seems to make the vectors wrong, that certain fragment shaders calculate, during their working their way across the bitmap…
Oops - this was in response to me - didn’t notice that at first… 
Don’t know what the render targets are today; Pimax have tweaked them multiple times, between updates, and I sold my 5k+ the other week, to offset the monetary hurt from an order for an 8kX, so I won’t be checking anew, but last time I went through all the modes and noted down the numbers, which was admittedly quite some time ago, I am pretty sure things did not remain proportional…
I think if nothing else, they may quite likely have taken a note from your research, and reined in the unnecessary vertical growth all the way up to the corner out in the peripheral hidden area mesh…
I read it somewhere in the SteamVR community threads months ago. They might have branched it back in for all I know
It’s definitely not a case of just exceeding maxRecommendedResolution (I do have the fix in place): Take Skyrim VR - I’m pushing enough pixels to the HMD that there is zero quality difference between the different FOV settings, as long as PP is turned off. Same is true with PP on in both Small and Normal FOV, but try to use Large FOV and there is a sudden and absolutely massive degradation in quality - we are talking like 1/4 the resolution. PCars2 displays this exact same behavior. So I’m basing my guess on only those two titles unfortunately. I’ll have to try AC as well and see if the same thing happens (although, AC seems to handle most of the VR settings itself - even the basic in-game resolution and AA settings function more or less as they should, so I’m not sure if testing that title will tell us much). All my other titles are through Vorpx.
Regarding 8KX and Large FOV, check my report thread:
https://community.openmr.ai/t/pimax-8kx-testing-progress-december-20-2019/24546/53?u=sweviver
Large FOV ( with PP enabled) looked like crap UNTIL I changed the MaxRecommendedResolution to 16384. You can use value 8192 which will make Large FOV without PP look much better and sharper than default. But for LargeFOV and PP enabled, the resolution is above 9000 horizontally, so 8192 is not enough.
Only problem is the performance. Most games run really good or perfect (on 2080Ti) in NormalFOV either with or without PP, but the Large FOV with PP enabled kills any GPU today if u keep SS 100%.
Having that said, Large FOV and PP enabled has never looked so good before.
Alright, I admit, I do not know, what are the default target res on the current PiTool either, as I am stuck with v132, because the newer ones have the famous stutter. But it is not difficult to figure out for anyone with the new version (who has not sold the headset 
).
Sounds like a idea to release a native 4K 140 fov hmd for a cheaper price.
I respectfully disagree. IMO, the last thing Pimax needs at this point is yet another headset.
