Have you a link who explain how these miracle is done...

DPI mean dots per inch... and to my knowledge, it is a fixed value... if a screen is able to have 2560 pixel ( dots ), nothing change when you use a lower resolution... the screen itself have always 2560 pixel ( dots )... the amount of DPI remain the same...

By example, my little 20.6 monitor have a have a DPI of 96 ( http://www.infobyip.com/detectmonitordpi.php )... that the monitor run at 1680x1050 or any lower resolution, DPI remain 96...

Thoumsin ...I think you'll find that what you're getting is 4 pixels per pixel....that's 4 pixels of 'screen resolution' to each pixel of system 'painting'.....in other words [as Brad mentioned] the system display resolution is exactly half of the hardware resolution...and thus twice the clarity/sharpness.

That makes it a step beyond 'native resolution' where system res is identical to screen res... 1 pixel per pixel ... and obviously better than non-default resolutions where the physical pixels 'fudge' [blur] the data displayed....particularly an issue with LCDs ...

What is see is that frame is rendered at lower resolution ( graphic card ) and send to a high resolution monitor where several pixel ( 4 in this case ) will render the same thing...

when is setup my windows OS to 800x600, my monitor remain at 1650x1050... everything become bigger on the screen, more physical pixel are used for render a single software pixel... but this don't increase the dpi...

It is only improving a little if your system is using old bipmap font or icon... but with the advent of vector graphics and fonts in place of bitmap graphics and fonts at the end of the last century, it decrease the antiliasing quality...

When i look at the video http://www.youtube.com/watch?v=jS-hBRqS4Tk , it seem to me that it is nothing more that a zoom feature... have the same effect on windows when i switch from 96 dpi to 120 dpi font/icon... everything become bigger on screen...

Well, until there is a real explanation about what really happen... i will simply consider it like a Apple marketing plot dusting some 10 year old technology, call it new and say that they have invented it...

No...4 pixels are available to render what one would at 96dpi...so there is a potential for 4 times the data/information to be displayed..... not 4 pixels with the same information....that's just magnification and thus pointless.

http://www.cambridgeincolour.com/tutorials/image-interpolation.htm 

Possibly that explains what I mean.....

 Drool some more. 3 x 30" displays. 

Thoumsin, I was having a hard time understanding this too, but I think I got it (?): zoom simply pixelates the result (i.e.; a single pixel becomes a 4 pixel bulky square), but here each 4x4 pixel is 'worked some more' - a bit like the difference between normal text vs. anti-aliased text.

Well, in fact, it perfectly explain what I mean...

Even the most advanced non-adaptive interpolators always have to increase or decrease one of the above artifacts at the expense of the other two — therefore at least one will be visible...Adaptive interpolators may or may not produce the above artifacts, however they can also induce non-image textures or strange pixels at small-scales...

Let sat that you have a monitor at 2560x1440...

In place of render a high quality frame at 2560x1440, the HiDPI render a lower quality frame at 1280x720 who is interpolated at 2560x1440 by software...

A original frame at 2560x1440 will always be more sharp and have more detail that one at 1280x720 who is interpolated at 2560x1440...

As one of your inhouse modeler/texturer how he will work for the best quality... let say that we need a 2048x2048 texture for a soase capitalship... what is the best way :

- make a 4096x4096 texture who is downsampled at 2048x2048 at the end

- make a 1024x1024 texture who is upsampled at 2048x2048 at the end ( HiDPI way )

Wish one do you thing will have the best quality with the more detail...

Well, it cannot be other... the iphone 4 screen is 640x960 for a 3.25 inch size... this give us 326 dpi ... who is more detailed that human eyes can see... same if some artifact is created during resampling, the artifact will be so small that a human eyes don't see them...

But for the same trick work on a 32.5 inch size, you will need a basic resolution at 6400x9600 for keep the same 326 dpi...

Well, there is nothing wrong with HiDPI in itself... will become very useful in a near future when home desktop will be UHDTV ready ( http://en.wikipedia.org/wiki/Ultra_High_Definition_Television )... more monitor is near the physical 300 dpi limit of human eyes, the trick work the best... and when it is over 300 dpi, it really improve the final result because the trick allow to use the extra dpi ( these over 300 ) that human eyes cannot see... more, the 300 dpi eyes limit is true in case of very good eyes ( 20/20 )... for people with not so good eyes, the dpi level drop down... for some people, the HiDPI trick can already lead to better quality at 200 dpi...

As for true retinal display, there is a good article here : http://prometheus.med.utah.edu/~bwjones/2010/06/apple-retina-display/  :

So, if a normal human eye can discriminate two points separated by 1 arcminute/cycle at a distance of a foot, we should be able to discriminate two points 89 micrometers apart which would work out to about 287 pixels per inch.  Since the iPhone 4G display is comfortably higher than that measure at 326 pixels per inch, I’d find Apple’s claims stand up to what the human eye can perceive.

Well obviously it doesn't increase the real DPI of the monitor.

But what it does do is use the 2560x1440 to grow everythign to what it would look like at 1280x720 but with a clarity that is unreal.

Well, some render engine use the subpixel rendering who can somehow triple the horizontal resolution...

1 white line on grayscale monitor

2 white line on grayscale monitor with antiliasing

3 white line on color monitor

4 white line on color monitor with chroma subpixel

5 same that 1 to 4 with less zoom

If i wrote about these subpixel thing, it is because it is somehow used by SubLCD ( http://www.oyhus.no/SubLCD.html )... and the SubLCD method look very similar to these new Apple HiDPI method :

The picture goes through SubLCD first, which shrinks it to half the size, but double the resolution.

Well, nothing wrong there... seem that Apple have invent the method 25 year ago, Microsoft have "stole" it ( 1998 ) and patent it until 2019... SubLCD ( 2007 ) have found a way for similar result without patent conflict... Apple make HiDPI... seem like a full circle...

It is how i think that HiDPI work... i can be wrong since until now, i have not find any article who explain how it really work at the technic level...

Will not surprise me that it was implemented in new MAC OS in planning of future desktop retina screen who have 300 dpi or more... when these screen hit the market, these HiDPI thing will reach his full potential...

Cheers Jorge!