Could be but that argument renders any TTL shots basically moot.
I do know that the Index shots look the way the Index does and I trust SweViver to have taken adequate shots of the 8KX.
The difference is remarkable.
4 Likes
Hi Martin. Youâre the reason I backed pimax and I value your opinion today as much as I did then. Thank you for the great examples of the 8KX.
5 Likes
Reading over a variety of forums. It seems pitool render multiplier works something like the pixel density setting from the old piplay.
Now back on track. Prior to updates in VR many used to use Reshade with sweetfx to sharpen games. Unfortunately changes in VR these from my understanding only work now in flat programs.
The blur in VR is normal and requires usually higher SS to compensate. The general census out there is a sharpening filter/shader pass is needed.
So maybe a Soft to sharp slider might be able to be implemented (some games have added a sharpening setting). Only thing is too much sharpening may increase aliasing(jaggies).
Yeah having a cold/flu sucks for sure. Rest up and know yes you can have an opinion; itâs a forum after all. 
6 Likes
@Heliosurge Have you looked at my spreadsheets yet?
1 Like
Agreed it comes down to Higher Res and Higher Res Panels. As we had discussed before Upsampling/SS cannot create details that are not there.
3 Likes
I have glanced over them yes a couple of times.
Then please consider this, as I posted earlier in this thread
PiTool at 1.25 and SteamVR at something like 100%, vs PiTool 1.00 and SteamVR at 125%, with VirtualDesktop supersampling very high, is the best demonstration that a blur filter is in place.
In both cases, the Total SR is exactly the same - 1.25x - before Virtual Desktop in-app supersampling which is set to something like another 1.6x.
This is strong evidence there is an actual blur filter in SteamVR, precisely to minimize RGB âsparklesâ around the edges of sharp high-contrast objects.
I think this more comes down to how each platform super samples. As stated earlier as well. Oculus platform and often Unreal Games do not lile PiRender set to high. In some games like Ethan Carter it makes the textures look very low res.
Now the good thing with VD is you can get it on both Steam and Oculus platforms to be able to compare effects of base settings and their SS methods.
Oculus seems to have a stronger base setting that doesnât need to increase pixel density. One youtuber mentioned with Oculus he keeps pitool at a low setting x0.5 or 0.75 and uses oculus and in game settings to optimize graphic quality.
Remember SteamVR has been like this prior to them releasing an RGB headset.
2 Likes
AFAIK, Oculus still does not support an ecosystem of overlay apps, and some apps will aggressively try to use it if installed. That is why I have been reluctant to install Oculus again.
How each platform super-samples⌠another way to put this would be that somewhere in the rendering/compositing pipeline, whether as a separate processing step or not, is an algorithm that has the effect of spatial domain low pass filtering - a blur filter.
The exact opposite of a spatial domain low pass filter is a spatial domain high pass filter - a sharpening filter!
This stuff gets a lot simpler when we all realize there are only a few mathematical effects that can be at work here, and everything would be better if they were implemented as discrete steps with options to turn them off.
EDIT: To be clear, I donât intend to be aggressive with this post. Just making a point that modeling the combined effects as individual terms in basic equations makes it a lot easier to pin down where the real problem is.
4 Likes
The simple truth though is games and programs seem to have greater performance in Oculus mode and usually do not require PP as many do when ran in SteamVR mode.
We saw a blurring of details in piplay after 192. Later on they introduced to the community a json file setting to increase pixel density but said it would. âIncrease pressureâ on the gpu. The result was a sharper higher quality image that we had become used to in 192.
It still took them sometime to get the overall quality equal to and later better than 192.
I havenât tried it as of yet. But I think someone said Ethan Carter works well on pimax in Oculus where as it is unfortunately broken on pimax in steamvr. Demonstrating some issues in SteamVR. (Where as it worked well on p4k - Maybe @century can test p4k on pitool with Ethan).
Everspace is another title that fails in Steamvr Mode. But without Oculus installed can be launched from steam desktop in OculusVR mode.
You might similarly do the same with VD.
2 Likes
Itâs a discussion. I am sure you understand the difference and are unlikely to take things personally. Debates and opinions are quite healthy way to brainstorm ideas if one can focus without pride getting in the way. 
That being said to pin things down you often have to eliminate what you can to evaluate differences. Much like running a game on Amd and Nvidia with similar settings to see how each performs.
4 Likes
I do not know what you call âTotal SRâ, but the supersampling in these two cases is different.
4 Likes
Total Super Resolution, so named to avoid any confusion with anything that has become a marketing buzzword subject to casual imprecision.
PiTool âRender Qualityâ Vertical Pixels (2700 at 1.25 Render Quality)
SteamVR Vertical Pixels (3375 at another 1.25x SS)
App Vertical Pixels (4219 at another 1.25x SS)
DIVIDED BY
2160 native vertical pixels from Pimax Vision 8kX
4219 / 2160 = 1.95 Total SR
In practice, I set SteamVR resolutions by vertical pixels, rather than âpercentageâ, using values just slightly above those calculated by my spreadsheet, because SteamVR does not allow exact resolutions to be set.
The supersampling in the two cases I mentioned, if PiTool, SteamVR, and VirtualDesktop are indeed performing supersampling, is in fact, exactly the same, excepting maybe a tiny difference due to not being able to set exact vertical resolution values in SteamVR.
1.25 * 1 * 1.6 = 2
1 * 1.25 * 1.6 = 2
2 Likes
If you can add post processing injection support in Pitool to that list SweViver for the engineers, then it would be much appreciated! A sharpening filter will be hugely beneficial to final image quality and very cheap performance wise. Not to mention all the other filters we could use. Like advanced AA etc.
Direct display mode breaks most post processing injectors like Reshade from working. But Pimax obvs have access to what happens in that closed off pipeline?
Even just a simple sharpening toggle would be great.
Cheers, and thanks for all the hard work! I can definitely see the difference in those images.
1 Like
Setting SteamVR supersampling to 1.25 means 1.25x more pixels (in total), not just vertical, so I really canât follow your calculation.
Could you record the recommended render target resolution (both vertical and horizontal), reported by SteamVR, for each step?
5 Likes
Vertical resolution is used because it is easier to standardize on - it does not change with FOV, and can be used with different headsets. By using vertical resolution, it is possible to use the same supersampling calculations for Pimax Vision 8kX as HP Reverb (at least if PiTool Render Quality is 1.0x).
As for the render target resolutions, see the spreadsheets, load calculations. Hover over the vertical resolution cells there, and read the comments. Exact SteamVR render target resolutions (both vertical and horizontal) are written as tested with specific applications.
1 Like
Aha⌠I had been assuming your âSRâ to mean Sample Rate (âŚwhich is an expression that Iâd personally like to see to mean âworldâ sample rate, independent from any HMDâs resolution and FOV, and measured in Pixels Per Degree at the spot on the viewplane that is perpendicular to the view point), and have been somewhat confused. :7
(And as risa2000 says: The equivalent of 1.25 PiTool Quality, in SteamVR SS, is not 125%, but 125% squared (EDIT2: âŚbut maybe the â1.25â, written format, is to make it implicit that that conversion has already been âbakedâ? :7).)
I have a hard time believing there would be an outright blur filter in SteamVR, no matter how good the justification. What I could believe, is that maybe SteamVR does not, after all, hand the full supersampled âoversizedâ image over to piserver, like I would have expected it to, when distortion correction is deferred to the manufacturerâs own driver, but scaled down to 100% first. That would introduce unwanted softening.
I can not imagine why that would be the case (EDIT3: well⌠I just did :P) (âŚand maybe there are (EDIT: âŚAPIâŚ) options for the manufacturerâs driver to receive either), but hey⌠I have been wondering about the compositing step: How does SteamVR handle the compositing of application- and overlay- images, when they are rendered at different resolutions?
There is also a SteamVR option to opt to use an older texture filtering method, which provides a sharper image than the newer one, at the cost of more aliasing, but I imagine that should really not have any effect other than in cases when SteamVR does the distortion correction⌠Oh well. :7
2 Likes
Whatever the percentage change to SteamVR SS is from PiTool Quality, I donât pay attention. Vertical pixels are a consistent measurement.
Yes, of course I realize that any increase in vertical pixels results in roughly squared as many pixels rendered, and of course that is âbakedâ into all my calculations.
SteamVR scaling down to 100%⌠this would not be a problem if the interpolation did not introduce some blur. There are a few interpolation algorithms to choose from - the types that minimize artifacts do so by low-pass filtering in the spatial domain.
In fact, applying a such a low-pass blur filter is so well supported by traditional engineering thinking, I would expect most well-educated developers to implement it out of supposed necessity.
Personally, I design radio systems to include such filters, so they donât pick up off-channel frequencies, but test/measurement equipment (which is more like our headsets) to make them optional. I really like that my DSO Nano scope doesnât have such a filter - its ~200kS/s can âseeâ things into the MHz (and GHz with a diode). But most other people seem to just put the low-pass filter on no matter what, to avoid getting any âspuriousâ readings.
In truth, bypassing the blur filter (however it is implemented), will cause us to âseeâ some things that are not real - morie patterns and RGB speckle. Before the Pimax Vision 8kX and Valve Index, these were severe enough to reasonably consider invalid behavior. Now there are a few use cases (Virtual Desktop, DCS World, Elite Dangerous, etc) for which some users will want to disable the blur.
2 Likes
@SweViver Excellent screen shots, thank you, it is a shame some of us can not get the focus point as clear as the images, the IPD will not go low enough, would love to have the 8KX as it looks good, unfortunately the index will look better for Me as it is in focus.
2 Likes
Weâre not mindreaders, Iâm afraid, and enough writing does not make the necessary adjustment, that one can not make any assumptions at all, that somebody would; But now that I hear that you do: Let me be the first to say that I, for one, do appreciate that you use a consistent format, that actually works mathematically, without requiring the reader to anticipate being required to do it themselves - the guesswork on intent goes both ways. :7
4 Likes