Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

chleuasme

macrumors 6502
Apr 17, 2012
485
75
Now is that fraction a touch to the left or right ? Is either really what the dev wants anyhow ? We don't know, since he's never had to deal with this case.
Again, from the dev's standpoint, would it make any difference if he wouldn't have to treat the screen differently than other iPads. It's still a 1024x768 pts (not pixels) screen.

In this case we have no 1:1 or 4:1 relation between pixels and points, but does it really matter? We're talking here about a 250+ ppi screen, the error is only a fraction of ~0.1 mm. Doesn't look like a problem for a touch device.

So again, what I was saying is:
The touch point values evaluations would be considered as a coordinate in a 1024x768 canvas.
The display is a black box, devs consider it as a 1024x768 canvas to design their UI (but can supply bitmaps at different size for pixel perfect rendering on specified screen). The OS use the available pixels, and it works fine as long as there is enough pixels and the ratio is still 4:3.
 

Big-TDI-Guy

macrumors 68030
Jan 11, 2007
2,606
13
What are those displays sitting on top of in the photo? I'm referring to the glossy black surface that appears to have bolts in the side of it.

It looks very similar to an open gantry laser engraver (how components are marked), or perhaps a laser scanner (scanning hi res 3-d models, part measurements). Most engravers have smoke extractors on the active plane, and I can't see any in that photo.

Edit: Pretty sure that's a gantry laser scanner / toolmaker bench.

http://www.yx17.net/en_asp/product.show.asp?id=728
 
Last edited:

KnightWRX

macrumors Pentium
Jan 28, 2009
15,046
4
Quebec, Canada
Again, from the dev's standpoint, would it make any difference if he wouldn't have to treat the screen differently than other iPads. It's still a 1024x768 pts (not pixels) screen.

In this case we have no 1:1 or 4:1 relation between pixels and points, but does it really matter? We're talking here about a 250+ ppi screen, the error is only a fraction of ~0.1 mm. Doesn't look like a problem for a touch device.

Then if you don't consider input, consider graphics scaling. 1.5 scaled graphics wouldn't look good at all. Nearest neighbor scaling works great for integer factors, other methods that would look good at 1.5 scaling like bilinear filtering or bicubic filtering would result in some blurriness added to images.

All in all, it's all these little details that just add up and make 1.5 not ideal.
 

chleuasme

macrumors 6502
Apr 17, 2012
485
75
Then if you don't consider input, consider graphics scaling. 1.5 scaled graphics wouldn't look good at all. Nearest neighbor scaling works great for integer factors, other methods that would look good at 1.5 scaling like bilinear filtering or bicubic filtering would result in some blurriness added to images.
Yep, clearly, and that's why I don't think this scaling-only solution is the right one. And this is not what we'll have.

But, look at the rMBP: you have a 3840x2400 resolution scaled on 2880x1800 pixels at 220 ppi, and the blurriness is really bearable. The high pixel density of the screen "makes" part of the work for your eyes.
You could imagine a similar situation with the 2048x1536 resolution rendered on an intermediary resolution with a high enough pixel count. And at a high pixel density, scaling is smoothed.
Well, that's clearly not happening, this would mean for the mini as much horsepower as the retina 9.7 iPad, with same CPU+GPU, batteries capacity, to be able to handle the same resolution.
That's why few months ago I tried to imagine an alternate solution, that I tried to discuss here or there. But that's clearly not happening, too :D
 

KnightWRX

macrumors Pentium
Jan 28, 2009
15,046
4
Quebec, Canada
But, look at the rMBP: you have a 3840x2400 resolution scaled on 2880x1800 pixels at 220 ppi, and the blurriness is really bearable. The high pixel density of the screen "makes" part of the work for your eyes.

Actually, having looked at it closely in terminal using a bright font on a black background, there is no blurriness at all. It seems Apple is not using a scaling algorithm that results in pixel interpolation (using a certain shade of gray to introduce or remove pixels between a white and black pixel), but rather in simply adding/removing pixels like nearest neighbor would do (but obviously not nearest neighbor, as that just makes bigger pixels out of smaller ones or simply deletes extra pixels).

It's obvious they're not using the same scaling solution that the public Quartz API exposes in CGScaleContextCTM(), as that is just an uncontrollable mess of blurriness (I think it's plain bi-linear filtering, no option for devs to even choose a different algorithm as far as I could find in the docs and online).
 

0092762

Cancelled
May 29, 2005
273
316
At 1.5 scale factor ? Touch points would be affected. You can't have 1024x768 touch points on a 1536x1152 screen without a few points falling between 2 pixels...

That's the problem with non-integer scaling factor.

Does it matter that much if the touch points are off by a single pixel? I doubt anyone's fingers are that precise. You can always analyze where most of the user's fingers are touching to determine the correct one as well.

Still, I think that's too complex and expensive and Apple will just use a 1024x768 screen for cost.
 

chleuasme

macrumors 6502
Apr 17, 2012
485
75
Actually, having looked at it closely in terminal using a bright font on a black background, there is no blurriness at all. It seems Apple is not using a scaling algorithm that results in pixel interpolation (using a certain shade of gray to introduce or remove pixels between a white and black pixel), but rather in simply adding/removing pixels like nearest neighbor would do (but obviously not nearest neighbor, as that just makes bigger pixels out of smaller ones or simply deletes extra pixels).
Isn't this specific to font rendering, with still the OS X font rendering engine taking care of the resolution change?
 

chleuasme

macrumors 6502
Apr 17, 2012
485
75
why didn't these guys examine their parts more deeply to know if it's retina or not?
That won't give the size of pixels, but playing with levels and contrasts in the posted picture, you can reveal what could be the real surface of pixels. It roughly appears to be 120 mm x 154 mm. Then smaller than 7.85" but enough for at most 1600x1200 pixels at 264ppi (screens usually don't use every pixels, leaving dead borders).
Fake?

 

guineapig

macrumors regular
Oct 5, 2011
169
0
Italy
That won't give the size of pixels, but playing with levels and contrasts in the posted picture, you can reveal what could be the real surface of pixels. It roughly appears to be 120 mm x 154 mm. Then smaller than 7.85" but enough for at most 1600x1200 pixels at 264ppi (screens usually don't use every pixels, leaving dead borders).
Fake?

[url=http://s8.postimage.org/jjfcfm8b9/eim.jpg]Image[/url]
I guess the only way to know is by pointing a microscope at the display and measure the physical size of a single pixel.
Or wait for the keynote :)
 

nofunsir

Suspended
Dec 30, 2009
83
53
Reno
16:9 is only good for watching movies. When you hold the iPad vertical it is rubbish. Especially for reading books, which will be a _major_ use of the iPad mini. On the iPhone 5, there is the positive that you can still cover the width with one hand, that won't work on an iPad mini so no redeeming advantage.

And there is one feature that sets it absolutely apart: Less size. No other difference is needed or wanted.




1024 x 768 means all software runs unchanged (just smaller). Pixel density would be higher than on the iPad 2, so you would have "almost but not quite retina" quality. 2048 x 1536 would be a waste of money. You also need more power with more pixels, and you don't have the space for a large battery on an iPad Mini.



Actually, 1024 x 768 would not be almost retina quality, especially considering the iPad mini would be held closer, and used heavily for reading. It's only a small increase from 132.0 DPI to 163.1 DPI.
 

Attachments

  • display_matrix2.png
    display_matrix2.png
    275.7 KB · Views: 157

bingo-champion

macrumors newbie
Oct 25, 2012
1
0
iPad Mini Scale Factor

Based on my analysis the iPad Mini will likely have physical screen dimensions of 160mm x 120mm and a scale factor of 1.25. iOS developers will be able to access the actual value on a physical device with:
[UIScreen mainScreen].scale

Near as I can tell the LCD screens in iDevices are constructed with Metric measurements (millimeters), but are advertised in Inches for greater US public appeal.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.