The (non) importance of PPI for VR

There you go

More technical.

It is actually same shit as PPI only sound fancy and like Apple invent hot water again.

OK. And how do you know that have real 2940PPI and high SDE in use?
So if it is true because of the display size need to be magnified 3x higher than “regular” VR display but still it is higher than 800PPI but not much more than 1000PPI if we translate masure after magnification. But this rase question of optic quality. This could have possible more distortion picture on periferal something like fisheye or if we use Fresnel lenses more problem with God rays.

We can’t know nothing before first proper reviews.

It’s actually really simple. Imagine you have the pimax display panel right ? Now shrink it half in size. You have now 1600 ppi right ? It’s still 4k resolution though ! So to display it you now need to magnify twice. The result will be the exact same image, yet the DPI of the panel is two times higher.

DPI says nothing because it’s a function of panel size and panel size is irrelevant to the user (as long as the lenses can “translate” (magnify) it correctly). You shrink a panel in half so you get two time higher PPI ? Well that’s great but now you need to magnify twice :slight_smile: So the net result is 0.

DPI is irrelevant. It’s all about panel resolution (and FoV).

Oh and what matters somewhat too is the amount of used pixels. For example, the Pimax uses soft IPD, so they move two circles around the display, according to the IPD: the higher the IPD, the more space between the two circles (one circle for each eye being displayed). Only those 2 circles get picked up by the lenses though ! So there are relatively a lot of pixels that are NOT used for the final resulting image. So effectively this lowers the resolution.

Now imagine you have a square 2k*2k panel like Kopin made. Put two of them in a HMD, one for each eye and use hardware IPD to move them: this doesnt cause any lost pixels. So this will also be better for the SDE and picture fidelity.

Of course there are still lost pixels though, since the screen is square and the lenses are round, so you still effectively have lower real resolution than the panel resolution.

But the loss is smaller when you have hardware IPD.

Or in a picture:


Only the yellow pixels will be used and the blue pixes are wasted ! So hardware IPD is a lot better for the SDE and image fidelity than soft IPD, but only of course if square display’s are being used !

Actually this raises the question: which form factor will the Pimax 8k panels have ? Because if they’re regular panel size, the ‘waste’ is enormous. In fact, you’ll get this:

So the resulting SDE will be a LOT higher on the Pimax 8k than on the Pimax 4k if the panels are not square, due to huge unused panel parts. UNLESS somehow special lenses can be used. Maybe the fresnel lenses overcome this problem and make more oval than circle projected parts ? Hmmm thinking of this, it almost must be that with special lenses they create oval images. Because if not, it would be useless to use 2 panels (because the size of the circles doesnt increase). So I’m guessing the Pimax 8k lenses work like this:

Would love to hear more about this @PimaxVR ! Or maybe @crony has some input ? But yeah I think it works like this and that this is also THE reason Pimax moved onto their own renderer, because the image ‘distortion’ is a lot different (oval instead of round circle) and they need a custom renderer/distortion to create this.

2 Likes

It is up to the HMD developer to choose what to do.
I prefer a square lens because I hate pixel waste.

2 Likes

You are mixing PPI with DPI. DPI is actually native resolution of display. In Pimax case it is 2560x1440 because 4K is not supported yet.
With some basic mathematics and knowledge of Pythagorean theorem you can calculate exact size of display and how much magnification you need for reach desired FOV.
That is why for example Microsoft MR have higher SDE compared to Vive/CV1 even have higher PPI but smaller display and need higher magnification which increasing SDE canceling PPI advantage.
IPD dont have nothing with SDE. Fact that with proper IPD you notice SDE better is because better and sharper picture in general with proper IPD.
Also 8K problem will be if they use standard lenses. Even Fresnel lenses, which will be implemented in 8K, are not perfect solution.
For VR we will need some special made progressive lenses which will be made for exact task needed in VR. Standard lenses which now use all VR are made for some general purpose not have im minde to fix some specific needs VR have. But that will increase significantly cost of the HMD.

I did mistype DPI when I meant PPI indeed. We’re talking about displays so we’re talking pixels, not dots :wink:

Yes, that’s indeed exactly my point. PPI says nothing in itself. If you make an extremely small 1k display panel, you have extremely high PPI but still only 1k in pixels available to create the HMD image, so horrible SDE. It’s all about resolution and the FoV; we should not consider PPI for VR as a valid parameter.

Please take a look again at the pictures above. In the very first picture there’s a lot of blue area between the 2 lenses, which are “wasted” pixels, They’re just being used a ‘IPD adjustment zone’ and not for display: if the user selects lower IPD, the displayed area’s (the yellow circles) move to the center and vice versa. So the actual pixels that are left to create the HMD image (the pixels inside the yellow lenses) are actually way lower. So that’s why soft IPD causes more SDE than hardware IPD (IF of course the panel size conforms to the lense size). Just compare the blue area in the first 2 picture’s, that’s my whole point.

I think you’re using an online translator right ? I think a part of my explanation got lost in translation. A last picture then to explain:

this is what happens when you adjust the IPD parameter, it just moves the circles around the display. But a lot of display pixels are unused for the HMD picture = worse SDE.

2 Likes

I don’t have time to debate about it with you but your assumptions are completely wrong and are not based on facts but wrong interpretation of some aspect. of VR. This debate is pointless when you don’t want to see what I’m talking about.
You are valuable member here on this forum and I respect that, in fact one of the most active with lot of good thing you done for community. Don’t get me wrong but with all my respect to you this time you are wrong.
Just answer one question, if you are right hows that that SDE is increasing with decreasing of PPI?

From what i understand he’s referring to 3 factors.

  1. size of panel
  2. ppi of panel
  3. amount of magnification to give equal fov

Example 5" panel with 800ppi vs dual 1.5" with say 800ppi per panel for combined 1600dpi.

So lets say 5" panel has focus 1 to get 100°Fov.
The dual 1.5" panels would require 2x focus/magnification to get 100°Fov.

If i understand correctly the sde effect would be equal in both setups.

Now with having a single 5" panel with soft ipd there is more wasted/unused area the panel vs dual 1.5" panel.

The dual panel will have less sde due to less screen waste in binocular vr.

1 Like

Someone needs to declare a new metric, Pixels Per Degree, or PPD. Might need a vertical and horizontal measurement.

1 Like

Ok let’s do it with an example then. My science teacher always said that a lot of times it’s easier to understand something when you take extreme value’s. So let’s do that:

Imagine an old school cellphone, one with tons of visible pixels, let’s say a display of 300x300 resolution, so extremely low resolution. Now imagine that with some new procedure someone can print such a display on an extremely small panel, say 0.01*0.01 inch. What PPI would such a panel have ? You’d need to calculate the diagonal panel length first with pythagoras: v(0.01 ^2 + 0.01 ^2) = 0.014 inch. Now enter the info in a PPI calculator like this: https://www.sven.de/dpi/ and you end up with a whopping 30304 PPI !! Now imagine that with some specific lens you can zoom this image enough so that you can show it in a VR headset. How do you imagine that this will look ? Remember that on the panel are still only 300x300 pixels available to make the image so you seen tons of pixels !

I’m really not sure how to explain this better. But PPI is a function of panel size and resolution. The smaller the panel, the higher the PPI (given the same resolution). But a smaller panel for VR just means you need to zoom more ! So PPI is irrelevant. What counts to the user is that he only has 300x300 pixels available to make up the image ! So it’s all about resolution.

Remember that we’re talking VR here, looking through a lens. If you look without a lens, sure, the higher the PPI the less pixels you see. But in VR, through a lens, you have to zoom to show a complete image to the user. So the smaller the panel, the more you have to zoom. And zooming means pixels become better visible.

2 Likes

PPI is pixel per inch. That inch is same size on 5" or 50" screen. That is the point
I explain it already but repeat now again and will use E3 and HP/ACER VR. Deepoon E3 have 2560x1440 resolution with around 550PPI on 5.5" size. HP/ACER VR got dual 1440x1440, 550PPI on 2.8" diagonal. Combined this two display give you 2880x1440, 550PPI 4.7", This 4.7 you can calculate using basic Pythagorean theorem. So PPI id same but even with better resolution HP/ACER to reachsame level of FOV and visible picture need higher magnified lenses. So if it is 20% higher magnification that increasing SDE as well for that 20%.
That is confirmed on testing where guy who was testing it and compare with CV1 say’s that why CV1 give less SDE.

He starts talking about SDE at 6:11 in the video. He says there’s more SDE “probably because it’s uniform throughout the entire lens, as opposed to the Rift and the Vive which they focus on the center to be the highest and that’s where it’s the least noticeable”.

So he blames the lenses. He doesn’t talk about PPI or did I miss a part ?

BTW, at 7:25 he talks about the a smoothing filter in the OSVR HDK2, which causes a blurring effect, which lessens the SDE. This also happens in the Pimax ! In the Pimax we have the shutter glasses, if you take them out the SDE becomes a lot more noticeable. So in the Pimax 4k it’s also a trick: less clear image = less noticeable SDE.

On diferent issue which I mention it as well before on asumption how could workVR on that small 3000PPI display. Higher magnification increasing blurriness out of center on peripheral. this is strictly lens problem and lenses don’t have nothing with SDE. Do you know what it SDE at all and how it is made?

Kopin is workin on special lenses for their 3000PPI panel headset. They obviously think it can be done, and I’m sure they’re right. We’ll see Q1 next year probably.

Sorry you’re losing me here, not sure what your point is, elaborate please.

Another way of looking at it is this: when you’re looking at a display through a lens, you don’t see the actual display. You see a ‘virtual image’, a version of the display that your lens created. Now the amount of pixels you have in ANY image, either real or virtual, is the main factor in SDE. So if you have 300x300 pixels in a real or in a virtual image (blown up to ‘real’ proportions), it’s still the same SDE.

1 Like

OK now when this is separate thread lets start from begining and determine what is what and how it is effecting on VR picture in general.
SDE or in full name Screen Door Effect is term used for effect your picture is look like you are watching true door with mosquito net on it (Screen door).
What is cause of this issue?
Any display is made from physically small squares separate with thinly black lines whot create this visible grid what creating SDE. That is the only reason for SDE. When we look at displays from sertan distance we can’t see this grid but when we come closer this grid start to be more noticeable. Density of this cubes is measured by the number of pixels one square can reproduced and it is determine in PPI. Higher PPI smaller square are and that is mean smaller grid.
VR is using optics to virtually enlage the screen so this enlargement is scaling up this squares and grid what it making it more noticeable.
How much SDE is noticable also effecting lenses by radial blur but this just masking SDE by decreasing quality of the picture.
To compare this best example is the Vive and CV1. You can notice on comparing VR’e SDE that Vive have better sharpnes but CV1 less noticeable SDE even got technically exact same screens. Point is CV1 using trick whot DK" users was hacking to reduce SDE effect by put plastic film on display which was blurring the picture and this blur was made this grid less noticeable.
Only thing what fiscally reducing SDE is higher PPI
Lenses effecting only on picture quality and problem of the lenses are radial blure, FOV and God rays.
IPD is effecting on sharpenes of the picture and 3D effect. Effect on SDE is minor and secondary.

SDE is strictly hardware issue and depend on display quality.

Due to size of display and needed magnification this 3000PPI is like 5.5" display with 900 - 1000 PPI but problem is lenses. You can’t reach quality with standard magnifying lens. They will need special made lenses for this purpose. Doable but this raising cost of the production so it is questionable are they reach desired results for affordable price. There is still too much question about this product and no real prototype which will answer on some of them. We need to see first testing. It is too early for judgement.

1 Like

Agreed. There’s a common relationship between the pixel size and the lines between that cause the SDE. So the smaller the pixel, the smaller those lines and the less SDE. We totally agree on that.

Again, we agree here.

Exactly. Less quality of the visible image = masking SDE.

Exactly. Higher PPI = smaller pixel = smaller lines = less SDE. We totally agree here.

What we disagree though (it seems) is that I say that if you magnify a display planel, then the pixels become bigger and the SDE too. So if you have a panel with very high PPI, but magnify it a lot, then the net effect will be 0. Hence, it’s not about the PPI of the source (the panel), it’s about the PPI of the virtual image that the lenses create. Very high PPI of a panel means nothing if the panel is so small that you need to magnify it very much. PPI of the panel only means something if you specify the panel size too (and the FoV). But it’s better to just ignore it and look at the resolution, that’s all you need to know (together with FoV)

Also look at it this way: what’s the size of the ‘virtual image’ ? It’s not expressed in cm’s or inches, but its expressed by the ‘FoV’ ! So if you know the resolution and you know the FoV you actually have what would correspond to the PPI of the ‘virtual image’.

1 Like