News for hackers. Stuff that matters.
Re:IMAX’s absurd attempt to censor Ars by murbard2 on Friday June 19, 2015 @11:21AM (#9745377)
1) No, the resolution is a function of the solid angle taken by the screen. Yes, there needs to be an uptick in PPIs but 8000x10000 is basically a hard cap on what's necessary (corresponding to a field of view of 135x180 with a visual acuity of one arc minute). There are already prototypes doing 2000PPI so it'll be at most a few years.

2) It doesn't have to be clear that VR will succeed for it to represent a threat to IMAX's business.

Re:IMAX’s absurd attempt to censor Ars by Retric on Friday June 19, 2015 @12:03PM (#9745642)
It's worse than that due to aliasing effects. Remember your rods are not aligned on a strict grid so need more than one pixel per rod. And, the eye contains ~120 million rods, so ignoring motion your already at ~40k x 40k per eye. Assuming you’re not moving they screen around to follow they eye or projecting on the retina 'resolution' would cap out at around 10 billion - 1 trillion pixels per eye.

However, clearly you can get away with far less than that. But, the point where increasing resolution is not noticeable at some level is ridiculously high.

Re:IMAX’s absurd attempt to censor Ars by JoeAltmaier on Friday June 19, 2015 @12:08PM (#9745682)
Confused: Sqrt(120M) is around 11000. Where did 40K come from?
Re:IMAX’s absurd attempt to censor Ars by Retric on Friday June 19, 2015 @12:13PM (#9745718)
Aliasing effects as you don't get 1:1 pixel per rod alignment. Currently we use 16xAA and Sqrt(16) = 4. So, 11,000 * 4 ~= 40K. Though the density is not constant over the eye so that's probably low balling it a little bit.
Re:IMAX’s absurd attempt to censor Ars by Someone1234 on Friday June 19, 2015 @11:31AM (#9745438)
I like with #1 how you start with the word "no" but then repeat back to me what I said. So we both agree, you need a higher resolution than we have today to clone IMAX in VR, and we won't get that unless VR takes off (as funding is needed to produce the screens). VR also has the weight X factor, how heavy is this prototype 2000 PPI screen? And much would one cost?

Both we agree on #1, we both said the same thing.

> It doesn't have to be clear that VR will succeed for it to represent a threat to IMAX's business.

I literally don't understand what it is you're trying to say.

If VR fails, it isn't a threat to IMAX business by definition. Something that isn't popular or doesn't exist cannot be a threat to something else.

Re:IMAX’s absurd attempt to censor Ars by tgb on Friday June 19, 2015 @11:59AM (#9745626)
No, you said it depends on virtual screen distance, mubard said it depends on solid angle. Mubard is correct and more relevant to the idea of "make the virtual screen larger and further away" which has an implied constraint where the solid angle is kept constant.

No one would argue that the coming gen vr devices are at the pixels per degree of IMAX, that's just wrong.

Edit: why are so many comments (particularly top level ones) down voted in this thread?

Re:IMAX’s absurd attempt to censor Ars by Someone1234 on Friday June 19, 2015 @12:08PM (#9745681)
> No, you said it depends on virtual screen distance, mubard said it depends on solid angle.

No, I said it cannot do virtual screen rendering at lower resolutions. mubard agreed. The virtual screen would look terrible as you pulled back if the resolution wasn't high enough to support it (which I don't believe it is, yet).

Re:IMAX’s absurd attempt to censor Ars by murbard2 on Friday June 19, 2015 @12:20PM (#9745793)
No you said:

"But the "further" away you project the virtual movie screen, the higher the resolution of the VR headset"

That is wrong, plain and simple. How far the movie screen appears in the VR headset has no impact whatsoever on pixelization and is purely a stereoscopic effect.

Re:IMAX’s absurd attempt to censor Ars by Someone1234 on Friday June 19, 2015 @01:49PM (#9746466)
> How far the movie screen appears in the VR headset has no impact whatsoever on pixelization and is purely a stereoscopic effect

We're talking about rendering pre-existing video at those distances, so it absolutely DOES cause pixelation as fewer pixels are used to render the video content than its native resolution.

A lot of people calling me wrong seem to be under the odd mistaken impression that we're rendering some generic object far away, but the discussion is about display pre-existing video footage at a distance.

If the VR headset is 1080p and the video footage is 1080p, if you put it far away (i.e. it only takes up 50% or 25% of the VR headset's screen real-estate) then it obviously cannot be rendered using all 1080p of the video's original pixels (since many are in use on the VR headset's display for non-video footage) and therefore a loss of detail is unavoidable.

The only solution to this is having a higher PPI than the source video. And as you move the video footage further and further away, the PPI needs to increase to keep the video footage rendered at its native resolution.

Re:IMAX’s absurd attempt to censor Ars by murbard2 on Friday June 19, 2015 @02:22PM (#9746658)
We're talking about rendering a more distant bigger screen, so the solid angle and the number of pixel used is constant. The point obviously isn't to give the appearance of a tv-sized image placed 100 meters away, it's to render the appearance of a kilometer-sized screen 100 meters away.

And if you're wondering: "but then what's the difference"? The difference is based on a number of perceptual clues, such as

- at what distance do your eyes have to focus (this one isn't handled by current generation VR displays and will be the trickiest to get)

- vergence: what the angle between your eyes is

- parallax between the screen and other objects in your field of view

Re:IMAX’s absurd attempt to censor Ars by Someone1234 on Friday June 19, 2015 @02:38PM (#9746759)
> We're talking about rendering a more distant bigger screen, so the solid angle and the number of pixel used is constant. The point obviously isn't to give the appearance of a tv-sized image placed 100 meters away, it's to render the appearance of a kilometer-sized screen 100 meters away.

If you stretch a fixed number of pixels you reduce the detail.

If you shrink a fixed number of pixels and display that at the same or lower resolution, you reduce detail.

In either case detail is lost.

Re:IMAX’s absurd attempt to censor Ars by murbard2 on Friday June 19, 2015 @03:16PM (#9747011)
I am genuinely puzzled as to what can create such a confusion. So to be very very clear:

If you were watching a 1000 pixels wide, 1 meter wide TV from one meter away, the level of details you perceive would be the same as if you were watching a 1000 pixels wide, 100 meter wide movie screen from 100 meters away.

The physical pixels on the movie screen would be larger, but you would perceive the same level of detail because they would touch the same number of rods on your retina.

If the two screens touch the same number of rods in your retina, how can you tell you're looking at a small screen upclose vs a large screen away? Several clues: parallax, vergence, and focus. VR can simulate two of those.

So if you were looking at the images produced by the VR set up for a large screen far away vs a small screen close by, they would occupy the same number of pixels on the VR display, the same! It would be just as detailed. But in one case, the two eyes would show you the images from two different angles, whereas in the other, the images would be much more similar. This is the only distance we're talking about, the perceptual distance created by the VR.

In all cases you're looking at the same number of pixels a few centimers away from your eyes, but perceptually it can be made to appear as if watching a large screen at a distance or a small screen upclose.

If you still don't understand, I can't help you beyond that point.

Re:IMAX’s absurd attempt to censor Ars by squeaky-clean on Friday June 19, 2015 @02:33PM (#9746725)
There's no such thing as "further away" with a VR headset. It's always an inch away, and the appearance of depth is just a stereoscopic trick.

> If the VR headset is 1080p and the video footage is 1080p, if you put it far away (i.e. it only takes up 50% or 25% of the VR headset's screen real-estate)

How far away it appears has nothing to do with the viewing angle, unless you also mention the virtual size of the virtual screen. A 10" screen at 1ft away will have a smaller viewing angle than a 1,000,000ft screen 500ft away.

Re:IMAX’s absurd attempt to censor Ars by Someone1234 on Friday June 19, 2015 @02:36PM (#9746742)
> There's no such thing as "further away" with a VR headset. It's always an inch away, and the appearance of depth is just a stereoscopic trick.

Meaning objects get smaller, meaning that they take up less space on the resolution, meaning that you need higher PPI to render them at the same level of detail.

> How far away it appears has nothing to do with the viewing angle

Nobody is talking about viewing angles. I certainly never brought it up.

Re:IMAX’s absurd attempt to censor Ars by squeaky-clean on Friday June 19, 2015 @02:38PM (#9746761)
> Meaning objects get smaller, meaning that they take up less space on the resolution

Why? It's a virtual screen, why wouldn't you scale it up as well as moving it further? The sky is really far away from me, but it's pretty huge in my field of vision.

> Nobody is talking about viewing angles. I certainly never brought it up.

Which is why we are all saying you're wrong. It's entirely dependent on viewing angle and not distance. Distance and virtual screen size are how you can determine the viewing angle.

Re:IMAX’s absurd attempt to censor Ars by dpark on Friday June 19, 2015 @12:16PM (#9745749)
It has nothing to do with "pulling back" the screen. If the physical display resolution is not high enough, the virtual screen will look terrible regardless of how "close" it is. Even if you were trying to show the pixels on the virtual screen by rendering it very "close", it would still look terrible, because the physical pixels would have insufficient resolution to render the virtual pixels uniquely and clearly.
Re:IMAX’s absurd attempt to censor Ars by dpark on Friday June 19, 2015 @11:55AM (#9745600)
He didn't repeat back what you said at all. You said this:

> "But the "further" away you project the virtual movie screen, the higher the resolution of the VR headset needs to be to support it."

He's saying that's not true. He's also right. The resolution the physical VR screens need is completely unrelated to "how far" you project the virtual screen.

Re:IMAX’s absurd attempt to censor Ars by Someone1234 on Friday June 19, 2015 @12:14PM (#9745729)
As things get smaller a greater PPI is needed to render the same level of detail. You don't mind losing detail, sure, you can keep the resolution the same and keep making it smaller and smaller (distance and size are one in the same in this scenario).

It ultimately boils down to "when detail gets smaller than a single pixel, it is lost." So the smaller the pixel the more detail you can render.

To be honest if you're watching non-HD resolution content, current generation VR headsets at the high end might be able to project it across the room and maintain detail. But HD? S-HD? IMAX? Meh, we aren't there yet.

And I agree with another comment that there is a "cap" on PPI. Once you reach it the human eye becomes the limitation, not the screen(s).

Re:IMAX’s absurd attempt to censor Ars by dpark on Friday June 19, 2015 @12:28PM (#9745882)
> As things get smaller a greater PPI is needed to render the same level of detail.

It really doesn't work that way. If your VR display is too low-resolution to convincingly render a virtual display 15 feet away, then it's too low-resolution to convincingly render the virtual display 15 inches away. Your physical display has the same resolution regardless of where the virtual screen is, and if it lacks sufficient resolution to render an object convincingly, it just cannot do it regardless of how distantly the object is rendered. To put it another way, your display has the ability to render a certain amount of detail, full stop. Where a virtual object is rendered is irrelevant.

Try this. Sit close to your monitor, close enough that text isn't crisp enough. Now zoom your display (Ctrl+'+' in most browsers), which is equivalent to bringing a virtual display closer. Note that the text still isn't crisp. If you could see the pixelation before, you can still see it. Moving the virtual display closer does nothing, because you aren't adding more resolution to the physical display.

Re:IMAX’s absurd attempt to censor Ars by Someone1234 on Friday June 19, 2015 @01:42PM (#9746418)
> To put it another way, your display has the ability to render a certain amount of detail, full stop. Where a virtual object is rendered is irrelevant.

But we're talking about a virtual display displaying pre-existing content (e.g. video footage). We aren't talking about the object itself.

If it is one foot from you in virtual space, each pixel of video footage might be mapped 1:1 into each pixel of the VR headset's screen resolution. Therefore the loss of detail is 0%. Every pixel of video detail is transposed onto the display in the VR headset.

If the virtual object is further way (e.g. six feet) it takes up a smaller percentage of the VR headset's display, and therefore fewer physical pixels on the VR headset's display are used to render it. So now instead of 1:1 you might be 1:4, and you're seeing 25% of the video's original detail actually rendered on the VR headset's display.

> Try this. Sit close to your monitor, close enough that text isn't crisp enough. Now zoom your display (Ctrl+'+' in most browsers), which is equivalent to bringing a virtual display closer. Note that the text still isn't crisp. If you could see the pixelation before, you can still see it. Moving the virtual display closer does nothing, because you aren't adding more resolution to the physical display.

Your analogy is backwards and doesn't apply.

But sticking to that same analogy, go get 1080p resolution image. View it on a 1080p display. You're seeing 1:1 (100%) of the detail within the image (all of it). Now zoom out, and again, and again. Detail has been lost. First a little, then a lot, and as you go too far back substantial amounts have been. This is because a 1080p image on a 1080p display can only display 100% of the detail when it is viewed at 100%. As the image is zoomed (or further away in the VR headset's case) it is not possible to continue to consume all of the details as the pixels don't physically exist to display it.

So solve this problem you need a higher resolution than the source material, and it gets higher and higher the more you zoom. If you viewed a 1080p image on a 8640p display, you could zoom out significantly with absolutely no loss of detail, because the pixel density is such that it can still render every dot within the image.

So in order for a VR headset to be able to render stuff far away with no loss of detail the resolution of the headset has to be much higher than the source video. The further away the video gets, the higher the resolution. Again, that is if you want absolutely no loss of detail (in the real world, some loss of detail is acceptable).

In order to play back a 1080p video on a 1080p VR headset, the video either needs to be right in front of the viewer (at 1:1), you'll have to accept loss of detail when moving it further away, or you'll need a higher resolution on the VR headset than the source video (e.g. 4x more pixels on the headset than the source video, when it only takes up 1/4th of the VR headset's display space).

If the virtual display-video is only 1cm squared on each of the VR headset's displays, then that 1cm needs to have the same pixels as 1080p to display 1080p video with no loss of detail. It is just physics. As you move things further away in virtual reality they get rendered by fewer pixels on the real-world VR headset displays, and therefore loss of detail is unavoidable. You can only counteract this with higher PPI than the source video (or just accepting the loss of detail).

Re:IMAX’s absurd attempt to censor Ars by dpark on Friday June 19, 2015 @05:14PM (#9747714)
> If it is one foot from you in virtual space, each pixel of video footage might be mapped 1:1 into each pixel of the VR headset's screen resolution.

If you can actually map pixels 1:1, you aren't in a virtual space anyway. Accomplishing this 1:1 mapping means locking the view into a single position. If you allow me to look around, then as soon as I turn my head, the angle skews and you're not mapping 1:1 any more.

> So now instead of 1:1 you might be 1:4, and you're seeing 25% of the video's original detail actually rendered on the VR headset's display.

If 1:1 looks good, then 1:4 rendered 4 times as far away looks exactly as good. If you can't see the pixels at 1:1 then you can't see the lost detail at 1:4. If you could see the lost detail at 1:4, it would mean the 1:1 image looks pixelated. Again, your physical display doesn't care about how far away something is rendered. It has a fixed resolution, and if it's too low to be convincing, it's too low to be convincing at any rendered distance.

Go play an N64 game (or find some screenshots). You'll see that everything in the foreground is pretty rough looking. Everything in the background is equally rough looking, because rendered distance doesn't change the display resolution.

> Now zoom out, and again, and again. Detail has been lost. First a little, then a lot, and as you go too far back substantial amounts have been.

This doesn't matter. If it looks good at 1:1, then it will look good at 1:4. If it doesn't look good at 1:4, it didn't look good at 1:1. The lost detail is irrelevant.

> In order to play back a 1080p video on a 1080p VR headset...

I'm pretty sure you just can't do this convincingly, no matter the "zoom" level. If you rendered the video taking the entire display, then it would unpleasantly fill your entire vision and it wouldn't be VR and it would look terrible anyway.

> that 1cm needs to have the same pixels as 1080p to display 1080p video with no loss of detail

You're stuck on the idea that you need 1:1 pixels. You don't. You need a display with sufficient resolution to be convincing to your eyes. Once you have that, you can render your virtual 1080p image at any distance that would work in the real world and not care about how pixels are mapping.