Great work! I spent a 2 hr flight trying to imagine a way to trick MSFS into culling based on the center of the two eyes. In case anyone is curious as I was: the only reason the culling occurs still on the left is that MSFS culls terrain based on the left eye’s projection “cone” - somehow switching it to the right resulted in less culling overall is that correct @mbucchia? I hope this work will show Asobo why its so dang important (and should be relatively easy for them) to support wide FOV HMDs! I still refuse to fly with blinders on!
Yes we’re using a cheap trick we found experimentally, where we have FS render the eyes swapped (we give FS the left eye pose when it expected the right eye and vice versa) and then we swap them back again when passed to the headset. This is very easy to do and gives good results.
We found a way to get rid of culling entirely but it always causes one eye to look distorted, and performing the “un-distortion” is very complex and costs a lot of time. So we settled for the cheap trick above.
There might be other ways we can improve the cheap trick but we havent found yet. Hoping to have some more time for this later.
I was looking for a way to take accurate screenshots with MSFS2020 today.
The screenshots in steamVR are affected by the PC monitor resolution and are saved at a reduced size.
I have tried enabling DSR from the nvidia control panel and upscaling my 4K monitor to 8K, but the screenshots still get reduced to half its size…
Also, steam screenshots consume additional rendering performance for monitor output apart from VR.
I have also tried OpenXR Toolkit screenshots.
I think it is the beautiful and closest lossless image to what I am seeing on the HMD.
It appears that only the main left screen is supported as that is a debugging feature.
I would have been very pleased if they added the ability to save the right eye, or both eyes.
Hey i actually already made that change for the next release!
Sorry for this little hijacking @all,
@mbucchia is there info about OpenXR and DCS with pimax by any chance? I did not find any specific info so since I saw your name here I am taking the chance.
Edit: created a topic for it DCS (Steam), Pimax and Performance Gains (OPEN XR )
One question - I can’t see the menu (it’s way off to the left). Is there a fix or a manual config file we can modify menu settings?
You can always run in flat-screen mode to change the menu settings. It’s a pain, but it works.
You’re most welcome mate!
We do appreciate your hard works and time spent on this project!
Would love to see more from you.
Please let us know if you require more support from us!
I took a screenshot of MSFS2020 running at 8KX,pitool2.0[NIS40%]9588x6320per eye.
It appears that steamVR limits the maximum rendering resolution to 8192.
Two reduced 8192x5399 images were connected.
The result is a huge 16384x5399 image for both eyes.
Running at about 25-30 FPS.12700k&3090
The reason I didn’t notice any difference between pitool 2.0 and 1.75 results may be due to the limitations of steamVR.
pitool2.0 9588x6320 , pitool1.75 8388x5528, limit 8192x8192?
I seem to be able to set a little more reduced drawing cost.
You can set / cancel this in an ini file. (also with PE, i mean)
But why do you drive such resolutions?
Even with this huge resolution image I am observing aliasing. I am trying to ascertain if it is a limitation of 8KX or a rendering artifact.
And I am still looking for what are the less than perfect rendering elements.
For example, the aliasing is noticeable on the instrument and exterior edges of the plane.
I tried NIS On/Off with the resolution intact and saw an improvement in aliasing.
It seems that at least in this case it is not yet the limit of the 8KX screen. I am looking for such factors for a better setup.
Aliasing is always found somewhere in digital imaging. That’s how it is with non-infinite frequencies.
The only (harware) remedy is to get the affected area smaller than half a device pixel.
Otherwise, only software helps - blur etc.
In fact, it seems more realistic and better for image quality to raise the pre-NIS base resolution with PITOOL 1.25 than with PITOOL 2.0’s supersampling.
It still seems that the parameters and results need to be studied carefully.
Been working all week with @coverman on and off, and I’m pleased to announce that OpenXR Toolkit will support the Pimax eye tracker to bring (dynamic) foveated rendering to Flight Simulator!
We’ve gotten a little bit late on the release, but I’m hoping that this version 1.1 will come out in the next 2 weeks.
Found these settings on reddit and its amazing! everyone should try this:
COLOR FIX: All credit to MRTV Discord Server User “Pepie”.
In the PiTool, do it like this,
Channel Blue: Contrast +5 / Brightness -4
Channel Green: Contrast +4 / Brightness -3
Channel Red: contrast +2/ Brightness -5
and at last
All channels: Contrast +1 or +2 / Brightness -5 or -4 and click apply
Isn’t amazing. Dark colors drown completely in black.
All - Contrast +2, Bright -2 or -3
Red - C +2, B -3
Blue - C +2, B -3
Pitool RenderQuality 2.0, steamVR100%, NIS:width35%height41%FFR:ON
I got almost stable 30FPS.
The image seems to be displayed more clearly than with RenderQuality 1.0 to 1.5.
By entering LensDistortio0 in UserCfg.opt, I get proper video without using devmode.
I would have liked the ability to fill with a single color of the average when culling, an additional feature to place the third FFR ring at the 120% position, and 1/32 in FFR.
SU9 caused the lens flare in the left eye to appear displaced from the position of the sun during NoPP. Probably a post-processing miscalculation due to the tilted display. Does anyone else see this?