Calm and civil discorse about update 31-07


#321

This comes from here:


in particular:

On the raw rendering costs: a traditional 1080p game at 60Hz requires 124 million shaded pixels per second. In contrast, the Rift runs at 2160×1200 at 90Hz split over dual displays, consuming 233 million pixels per second. At the default eye-target scale, the Rift’s rendering requirements go much higher: around 400 million shaded pixels per second. This means that by raw rendering costs alone, a VR game will require approximately 3x the GPU power of 1080p rendering.

They basically say that for 233 MPix/s at display res, they have to render 400 MPix/s, which gives 400 / 233 = ~1.7


#322

Itz is pointless to debate about native and SS resolution and even compare Vive, Rift, and Oddisssay with Pimax8K. Vive, Rift, and Oddissay have approximate same FOV of 110° and Pimax8K or 5K got 200° FOV.
So there are multiple factors which determine the quality and after all impression in VR usage. 3 major factors which effecting on itch other and creating final VR impression are the Native display resolution, PPI what we could call display hardware resolution and effect on SSD directly, and PPD what we could call end resolution what represent what you can see in the final in VR usage.
There is the 4th factor which also affecting on end VR impression make difference only between Vive/Rift/Oddissay vs Pimax 8K/5K and that is lenses, not just because of 110° vs 200° FOV it is also affecting on PPI and PPD because of the different magnification factor.
This 4th factor take it with little grain of salt because it is based on some assumption and maybe not fully correct and that how much really is different and if it is at all we will see only after first serious comparison after lifting NDA.

To not now repeat all this year’s discussion about displays resolutions, PPI, and PPD I will just for example post the famous VRMAN66 YT video with a comparison of the 3 different displays like we have today and are used in 5K and 8K Pimax.

Deepoon E3 is using the same PPI as 5K and 4K same as 8K so that is like difference should look like between 5K vs 8K. To get the picture.
this is all comparing under same lenses magnification under 110° FOV VR. Similar results you will get comparing newer WMR’s or Vive PRO.
Now I’m coming to that 4th factor, the lens magnification. More ao less all 110° FOV VR have a similar physical size of the displays and use similar magnification on lenses. So this is not changing factor if you are comparing only 110° FOV VR’s but 5K and 8K are 200° FOV VR’s. Let’s compare it in numbers:
Vive/Rift - 100° per eye / 1080 pixel wide per eye = 10.8 PPD, PPI = 430
DeepoonE3- 100° per eye / 1080 pixel wide per eye = 10.8 PPD, PPI = 520
Pimax4K - 100° per eye / 1280 pixel wide per eye = 12.8 PPD, PPI = 800
Pimax5K - 150° per eye / 2056 pixel wide per eye = 13.7 PPD, PPI = 520
Pimax8K - 150° per eye / 2056 pixel wide per eye = 13.7 PPD, PPI = 800
WMR/Vive PRO - 100° per eye / 1440 pixel wide per eye = 14.4 PPD, PPI = 520
Pimax5KX- 150° per eye / 2560 pixel wide per eye = 17.1 PPD, PPI = 800

So compare this pictures and check the differences in numbers so you can imagine how much better could be a picture before real comparison after NDA lift.
But where is here magnification or 4th factor you are talking about? Yea, we are not finished yet. As you can see major differences are not just PPD but PPI and most of all FOV. This is maybe is not an exact value but we will use it to get the point. To immerse inside VR we need lenses to magnify picture and stretch it to this 110° so it magnifies it, let’s say 3x. to get a picture about SDE and put value for it in numbers will be if we divide display PPI with Lens magnification factor. So let’s do some SDE comparison in numbers using this values and you will see how this video confirming this numbers.

Vive/Rift - 110° FOV, PPD = 10.8, PPI = 430 / 3x magnification = 143 SDE
DeepoonE3 - 110° FOV, PPD = 10.8, PPI = 520 / 3x magnification = 173 SDE
Pimax4K - 110° FOV, PPD = 12.8, PPI = 800 / 3x magnification = 266 SDE
Pimax5K - 200° FOV, PPD = 13.7, PPI = 520 / 2,2x magnification = 236 SDE
Pimax8K - 200° FOV, PPD = 13.7, PPI = 800 / 2,2x magnification = 363 SDE
WMR/Vive PRO - 110° FOV, PPD = 14.4, PPI = 520 / 3x magnification = 173 SDE
Pimax8KX - 200° FOV, PPD = 17.1, PPI = 800 / 2.2x magnification = 363 SDE

You can notice that on 200° FOV VR’s have a lower magnification factor than 110° FOV VR’s. that is to keep proper aspect ratio in 150° FOV per eye because of display ratio and FOV ve need different magnification. Less magnification means less SDE and increasing size of VR pixels what will result in additional increases of the picture quality.
All these pixels calculations, SS, and other stuff are pointless because VR is based on individual impressions based on multiple factors combined in general picture not just on individual technical parameters. SS is used in certain titles to increase more sharp picture and details needed to play such title but needs for SS is depended from title to title not related to VR hardware its self.

Now you can see what is my prediction based on well known pieces of information and assumptions we got right now and I won’t debate about it because no one can confirm it right now and will be interesting to see first serious reviews which will tell us am I right or not.


#323

I see, thanks! A problem kind of coming in from the opposite end, to the matter from days ancient, of as closely as possible matching live action film grain in CGI composite layers… :9

Aha, ok; It all checks out then: square root of that roughly equals 1.3 supersampling, which is indeed lower than the Vive’s 1.4 (…or 1.96, if going by pixel count rather than scale. -Hopefully at some point everybody can agree on a single way everybody counts it everywhere, so that confusion is eliminated - even should that way turn out to be the wrong one. :9)

(Hmm, 'seems in SteamVR, defaults for the CV1 is different horizontally and vertically, with 1.24 on the X axis, and 1.33 on Y… :7)


#324

I believe industry agreed on factoring simply the total amount of pixels, in other words, supersampling of factor 2 means rendering twice as much pixels, or in yet another words, using 2 pixels for 1 output pixel. Valve created confusion when they started using square root, possibly thinking that it is more intuitive (for them :slight_smile: ).


#325

It depends what you are actually discussing. For what concerns supersampling factor, there is one direct implication, which is the rendering power needed to draw the required supersampled image. This is quite important characteristics (for me the most important one) for the customer.

Second, while it is “all about perception”, there are clear physical limits and requirements, for which SS is just a mitigation. Everybody agrees that with higher SS the image is “better” (to certain extent).

Concerning PPD/PPI discussion, I would not rather go there, as it is even more complex for my head to wrap around, but if you are interested, I would recommend you reading the latest post from doc-ok at http://doc-ok.org/?p=1694 which shows that PPD is the function of polar coordinate and that the whole PPD/PPI thing is not that easy. One consequence is that perceived picture quality from camera shots through lenses may be misleading if the pictures are not cut from the same PPD regions.


#326

So basically it has become a phallic measurement analogue where in the end it’s the quality received that’s more important than the raw size available


#327

Yes and no. Depends, in general, this is it but from my calculation and some experiences with different VR’s we got. Part of my assumption and calculations are proven by the video I publish or some other similar you can find on YT.
Point is if we use all these factors Pimax5K or 8K will need equal CPU/GPU power and that will be, after all just around 20% higher max than For WMR or Vive PRO for example and that is just because of 200° FOV. Another aspect lowering needs for SS so after all basic higher quality and less need for SS brings similar hardware consumption. Vive and Rift are even lower basic quality what increasing need for higher SS to compensate imperfections of basic performances. As you can see by calculations we could expect in general better experience in both Pimax5K and 8K against Vive PRO. Same tracking quality, minimum less PPD but significantly better SDE and about immersion based on FOV it is out of any discussion better. And for all of that for just minimal better GPU.
For this saber something game or that I will call it VR Space invaders or Pavlov GTX 1070 or RX 580 will be good base GPU but for more serious games like for example Asseto Corsa, Project Cars 2, Elite Dangerous or Il2 BoS GTX 1080 or even 1080ti are needed for pleasant playing. DCS World will need 1080ti or even some new after release 1180 or 1180ti when it will be released.
Don’t get me wrong but this is if you don’t want to compromise too much. Less powerful GPU will be possible but with significant reductions of details.


#328

I did not really believe in the brainwarp promise from the outset, which is why this will not disappoint me at all. It is a pretty simple principle to skip a frame per eye and only feed one eye per Hz, so I thought that this will be something others like Valve and Oculus tried before and found to not be performant enough to implement. It may work in certain cases, or at least make the 8K usable where it otherwise would be completely unusable, who knows. But I suspect it in any case comes at quite a decrease in perceived comfort or smoothness, so that you will think twice about applying it even if you otherwise cannot play the game at all…


#329

Maybe they did (source would be appreciated, should you stumble over it), although I had never, ever seen that practice, before Valve changed the setting in SteamVR to it, and have yet to see it anywhere else; Then again, my sample size is small and not necessarily particularly recent. :7

I have other objections, but will try to keep them to myself. :7


#330

I see you guys worried about SDE but I had a 4k that had very low one and SDE on the 8K is really the last thing Iam worried about since it was noticable mostly on menus or when reading.
Best will be the field of view ^^


#331