I would like to start a discussion on the topic of resolution on Apple displays. Personally I think that Apple's line of computers have screens with too low pixel density.
I could not agree more. Currently, I am not buying into any 'book since the screen resolution bugs me. 1440 on the 15" would be my personal threshold, 1600 preferred.
Here, at the company, I had the chance to see one of the Dell 15" widescreens with 1920 resolution. Although the panel and background illumination are shitty, the screen real estate you win is just "wow". If you up the font size a bit, it is *so much* sharper than that of my ageing Titanium, its just a shame.
Apple, for christs sake, offer a 1600 resolution as a BTO, I'd gladly pay $500 more just for the better resolution.
This can't be true if you take the books into account. The 12" has 1024x768, the 14" does too.
Sorry, I meant to say Apple's desktop displays. Apple's portable display line-up is just confusing.
Yes. Pull through and modify Quartz Extreme to drop the notion of a screen pixel but treat all pixels as virtual pixels and simultanously offer a new, CoreGraphics based API for those applications that know about cm, inches and so on. Only this way can Apple hope to use 300DPI displays when they get cheaper.
Why even have two systems? Just have Quartz Extreme change the units of the Carbon and Cocoa co-ordinates system from 1px to 0.25mm (current desktop pixels). Instant resolution-independance. Text would just be drawn as vectors in mm and the Quartz window manager would multisample all the graphics for the actual display.
Barto
PS Don't be afraid to point out anything wrong with what I've described
Why even have two systems? Just have Quartz Extreme change the units of the Carbon and Cocoa co-ordinates system from 1px to 0.25mm (current desktop pixels). [/B]
This would cause problems on Power/iBooks as their "native" resolution is not 0.25mm/px but 0.20 or less.
If you move from the unitless screen pixel model to a model that takes physical units into account, you absolutely should take the opportunity and make sure the units are accurately rendered, else you create even more confusion.
Would it not be possible to have vector representations of all the GUI rendered as bitmaps with rules defining how the pixels should behave depending on the resolution set in the machine. These bitmaps could then be cached and put into temp memory for when the GUI needs them.
They would only need to be recalculated when the resolution changes on any given screen. For most people, escpecially LCD users, I doubt they change resolutions that often.
This would cause problems on Power/iBooks as their "native" resolution is not 0.25mm/px but 0.20 or less.
If you move from the unitless screen pixel model to a model that takes physical units into account, you absolutely should take the opportunity and make sure the units are accurately rendered, else you create even more confusion.
Ahem, no. Maybe you want an interface where you have some windows rendered per pixel and some measured in mm, but I don't. I don't know if it's even feasible.
Some figure has to be chosen to move per pixel interfaces to rez-independent, and .25mm may as well be it because that is what interfaces are built for in Mac OS X.
Take the next step, why not keep the co-ordinates system and just say instead of 1 = 1px, 1 = .25mm? That way you only get minor compatibility problems, limited to the graphics field where stuff is updated early and often anyway.
Would it not be possible to have vector representations of all the GUI rendered as bitmaps with rules defining how the pixels should behave depending on the resolution set in the machine. These bitmaps could then be cached and put into temp memory for when the GUI needs them.
That would only work if the boundries of the windows were per-pixel, because otherwise the detail of the window will change as it (the window) is more xmm left, right, up or down. And I don't think a per-pixel window boundry would work, without serious glitches. Imagine every application behaving like Internet Explorer.
Certainly an entire window could be cached as a very high-resolution bitmap, but I think (anyone? Programmer?) that would be pointless with modern graphics cards able to multisample 2D and 3D graphics so well.
That would only work if the boundries of the windows were per-pixel, because otherwise the detail of the window will change as it (the window) is more xmm left, right, up or down. And I don't think a per-pixel window boundry would work, without serious glitches. Imagine every application behaving like Internet Explorer.
Certainly an entire window could be cached as a very high-resolution bitmap, but I think (anyone? Programmer?) that would be pointless with modern graphics cards able to multisample 2D and 3D graphics so well.
Barto
Im sorry, I dont really understand your reply, as im not a programmer,
what Im trying to say, is that (as I understand it), say a gui widget is 32x32 pixels, but you might want the same widget at 40x40 or 64x64 depending on the resolution of the monitor, or the pixel density.
I understand that you cant simply scale a 32x32 up to 40x40, because say for instance that there was a 1 pixel black-line border in the 32x32, it would become 1.15 pixels or something, and look bad.
So I think, a vector description of the widget, with "rules" that this black line can only be rendered as 1 pixel up to 63x63 or as 2 pixels at 64x64 or above, say for instance there is a red circle in the centre of the widget, a rule saying that the circle is always centred and its outline is always 1 pixel or whatever.
Then the OS would render out all the widgets as bitmaps, and the OS would call on these to display the GUI.
I dont see why its not possible, but perhaps you could elaborate your answer in the fashion of "GUI rules for dummies!" so I understand
You can either do something per-pixel or resolution independant. What you want is to have it both ways. Everything would be slightly out-of place, so you'd lose the accuracy of the graphics. The graphics would be single-sampled, so it would look like Mac OS 9 through a nearest neighbour resize-filter. Trust me, it would be bad.
You can either do something per-pixel or resolution independant. What you want is to have it both ways. Everything would be slightly out-of place, so you'd lose the accuracy of the graphics. The graphics would be single-sampled, so it would look like Mac OS 9 through a nearest neighbour resize-filter. Trust me, it would be bad.
Barto
PLEASE, explain to me why everything would be slightly out of place. Im thick, I just do not get it.
You need a 32x32 graphic at a certain place, you have a rendered 32x32 graphic from a vector discription, you place it at the right co-ordinate, how can it be out of place?
With a mm based rather than px based co-ordinates system, when a window moves it moves by maybe 1.15 pixels (to use your example) at a time. If one scaled bitmap or vector moves by 1.15 pixels left, then two pixels left will inherit .15 of that scaled bitmap, colour or vector. This will occur for every pixel the image is displayed on.
This is A Good Thing&trade, as you get more accurate display (if a tiny bit fuzzy). Think of the text rendering in Mac OS X currently, but for all the windows and graphics. The biggest advantage of this is the resolution independence of the system, and it will look better and better as the resolution of displays increase.
What you describe is to have co-ordinates in mm, but the boundries wrapped to the nearest pixel. That would certainly allow caching of object textures, but the edges of objects would appear closer or futher away depending where they are positioned on screen, even if the user specifies them to be the same distance (in mm). The edges of objects would also be aliased, unless you applied some kind of "smooth type" effect between pixels of the same object border.
It is certainly do-able, but there are disadvantages in the system for not much gain.
[This is a repost of something I just wrote over at (Macrumors) Mac Forum.]
When I first learned that Apple was expected to switch to a 15.4" display on the expected new Powerbooks (up from 15.2") I was shocked to learn that this increace in screen size would actually mean a decrease in screen resolution. Rumors claimed that it would go from 1280x856 (15.2") to a resolution of 1280x800 (15.4"). This in my mind is just absurd. Dell also delivers a 15.4" laptop these days and offers a choice of two resolutions for it. The high-end has a crisp 1920x1200 WUXGA screen and the low-end offers a measly 1280x800 WXGA screen-resolution.
If the rumors are true (either for this Powerbook update of a future one) that Apple would downgrade its display on their professional line of computers so that it would match that of Dell's budget variant, I might have to think twice about making the switch I have so long yerned to make.
Personally I think that there must be middleground to find here, because while I find the current resolution of 1280x856 unsatisfactory the 1920x1200 resolution offered by Dell may, for some people, be too much. Surely Apple could offer the high-end WUXGA display as a BTO-option or use a resolution that is higher than today yet not quite as high as the one on the high-end Dell portable. Personally I hope that the last of the two options is the way forward for Apple's 15.2" (possibly 15.4") Powerbook.
1280x800 can be very crisp. This obsession for as Amorph put it once, "Squintronic" screens leaves me wondering. There is no practical benefit unless the UI is up to it. Even then, the plethora of badly done web pages, like that squinty text garbage KLD is so proud of, will keep your nose 2 inches of the screen for a good long while.
If Apple goes with a 16:10 15.4" screen, their options would be 1280x800, 1440x900, and 1920x1200. Those are the screen resolutions we know exist at that size. In the case of a laptop, I'm generally willing to tolerate a little more dpi, because of the portability and mostly because the screen sits closer to your face most of the time, and a higher dpi is more tolerable on a crisp LCD than on a CRT display.
Apple could go with 1440x900, the same as the 17, and bump the 17 up to 1600x1024, if such a panel is available.
The availability of panels though is the key. For comfort, the dpi of the 1280 (15.2/15.4) and 1440 (17) widths is well judged.
I too think that Apple's current display resolution offerings are not high enough (but folks like it or not if Apple thinks it's enough, it's enough). When people argue that they are 'just fine' they always seem to compare them to the USXGA extremes that PC companies OFFER. The difference is, Dell and friends offer OPTIONS with the displays. XGA for the blind, SXGA who want it crisp, but without squinting, and USXGA (or whatever letters they use) for the super tight 'must have the highest resolution' freaks. It's all about options and what the individuals who are making the purchase prefer. Apple doesn't do options well they tend to offer extreme configurations with minimal choices. Also, for all of your 'friends' who've ordered Dell's and HATE the super high resolution, did you ask them why the hell they ordered it like with that option (at a premium) in the first place? Also, why didn't they just take advantage of Dell's 30 day satisfaction guarantee return policy and return it for a credit and exchange it with a preferred resolution? That's like buying a wide screen HDTV and wondering why the picture looks like a letter box or 'why is it so short and wide!?'
For the 15.4" PowerBook (if it EVER gets here), I'd like to see a resolution somewhere around the 1440x900 area like Matsu suggested. Since I won't pretend like I know what the rest of the Mac buying world wants or should need like some others here, I hope it's an Apple option. More importantly, I hope that Apple uses better displays than on the 17" model which looked a bit washed out on several display models that I've seen (like the iMac). Does anyone know who the OEM on those displays are?
1280x800 can be very crisp. This obsession for as Amorph put it once, "Squintronic" screens leaves me wondering. There is no practical benefit unless the UI is up to it.
At least for me, part of the problem is Aqua. Since they upped the standard font size from 9px (MacOS 1-9) to 12px, the 1152 pixel of my TiBook display roughly as much text across as did the 800x600 disply in my Wallstreet. As a coder, I need to have a lot of text on the screen simultanously.
Next is the Quartz antialiasing. It sure looks nice from a distance, but on a notebook display it is just mushy imho and the color-seams at the edges of characters bug me too. With smaller pixels, the effect should be less annoying.
Quote:
] Even then, the plethora of badly done web pages, like that squinty text garbage KLD is so proud of, will keep your nose 2 inches of the screen for a good long while.
HAha, doesn't work for those fixed ratio flash pages that some sites insist on using. Yes flash might be cool for a few effects, but it should be limited to ONLY certain effects and presentations and otherwise be part of a good old HTML page. At the end of the day legibility is more important. What's sad is a bunch of web designers and their clients think these pages look better and try to cram a bunch of small text into them, because they'd rather have that page than have it as a comliment to a nice legible HTML.
Comments
Originally posted by KANE
I would like to start a discussion on the topic of resolution on Apple displays. Personally I think that Apple's line of computers have screens with too low pixel density.
I could not agree more. Currently, I am not buying into any 'book since the screen resolution bugs me. 1440 on the 15" would be my personal threshold, 1600 preferred.
Here, at the company, I had the chance to see one of the Dell 15" widescreens with 1920 resolution. Although the panel and background illumination are shitty, the screen real estate you win is just "wow". If you up the font size a bit, it is *so much* sharper than that of my ageing Titanium, its just a shame.
Apple, for christs sake, offer a 1600 resolution as a BTO, I'd gladly pay $500 more just for the better resolution.
Sorry, I meant to say Apple's desktop displays. Apple's portable display line-up is just confusing.
Yes. Pull through and modify Quartz Extreme to drop the notion of a screen pixel but treat all pixels as virtual pixels and simultanously offer a new, CoreGraphics based API for those applications that know about cm, inches and so on. Only this way can Apple hope to use 300DPI displays when they get cheaper.
Why even have two systems? Just have Quartz Extreme change the units of the Carbon and Cocoa co-ordinates system from 1px to 0.25mm (current desktop pixels). Instant resolution-independance. Text would just be drawn as vectors in mm and the Quartz window manager would multisample all the graphics for the actual display.
Barto
PS Don't be afraid to point out anything wrong with what I've described
Originally posted by Barto
Why even have two systems? Just have Quartz Extreme change the units of the Carbon and Cocoa co-ordinates system from 1px to 0.25mm (current desktop pixels). [/B]
This would cause problems on Power/iBooks as their "native" resolution is not 0.25mm/px but 0.20 or less.
If you move from the unitless screen pixel model to a model that takes physical units into account, you absolutely should take the opportunity and make sure the units are accurately rendered, else you create even more confusion.
They would only need to be recalculated when the resolution changes on any given screen. For most people, escpecially LCD users, I doubt they change resolutions that often.
???
Originally posted by Smircle
This would cause problems on Power/iBooks as their "native" resolution is not 0.25mm/px but 0.20 or less.
If you move from the unitless screen pixel model to a model that takes physical units into account, you absolutely should take the opportunity and make sure the units are accurately rendered, else you create even more confusion.
Ahem, no. Maybe you want an interface where you have some windows rendered per pixel and some measured in mm, but I don't. I don't know if it's even feasible.
Some figure has to be chosen to move per pixel interfaces to rez-independent, and .25mm may as well be it because that is what interfaces are built for in Mac OS X.
Take the next step, why not keep the co-ordinates system and just say instead of 1 = 1px, 1 = .25mm? That way you only get minor compatibility problems, limited to the graphics field where stuff is updated early and often anyway.
Barto
Originally posted by Anna Mated
Would it not be possible to have vector representations of all the GUI rendered as bitmaps with rules defining how the pixels should behave depending on the resolution set in the machine. These bitmaps could then be cached and put into temp memory for when the GUI needs them.
That would only work if the boundries of the windows were per-pixel, because otherwise the detail of the window will change as it (the window) is more xmm left, right, up or down. And I don't think a per-pixel window boundry would work, without serious glitches. Imagine every application behaving like Internet Explorer.
Certainly an entire window could be cached as a very high-resolution bitmap, but I think (anyone? Programmer?) that would be pointless with modern graphics cards able to multisample 2D and 3D graphics so well.
Barto
Originally posted by Barto
That would only work if the boundries of the windows were per-pixel, because otherwise the detail of the window will change as it (the window) is more xmm left, right, up or down. And I don't think a per-pixel window boundry would work, without serious glitches. Imagine every application behaving like Internet Explorer.
Certainly an entire window could be cached as a very high-resolution bitmap, but I think (anyone? Programmer?) that would be pointless with modern graphics cards able to multisample 2D and 3D graphics so well.
Barto
Im sorry, I dont really understand your reply, as im not a programmer,
what Im trying to say, is that (as I understand it), say a gui widget is 32x32 pixels, but you might want the same widget at 40x40 or 64x64 depending on the resolution of the monitor, or the pixel density.
I understand that you cant simply scale a 32x32 up to 40x40, because say for instance that there was a 1 pixel black-line border in the 32x32, it would become 1.15 pixels or something, and look bad.
So I think, a vector description of the widget, with "rules" that this black line can only be rendered as 1 pixel up to 63x63 or as 2 pixels at 64x64 or above, say for instance there is a red circle in the centre of the widget, a rule saying that the circle is always centred and its outline is always 1 pixel or whatever.
Then the OS would render out all the widgets as bitmaps, and the OS would call on these to display the GUI.
I dont see why its not possible, but perhaps you could elaborate your answer in the fashion of "GUI rules for dummies!" so I understand
Thnx
You can either do something per-pixel or resolution independant. What you want is to have it both ways. Everything would be slightly out-of place, so you'd lose the accuracy of the graphics. The graphics would be single-sampled, so it would look like Mac OS 9 through a nearest neighbour resize-filter. Trust me, it would be bad.
Barto
Originally posted by Barto
Anna Mated, that would look horrible.
You can either do something per-pixel or resolution independant. What you want is to have it both ways. Everything would be slightly out-of place, so you'd lose the accuracy of the graphics. The graphics would be single-sampled, so it would look like Mac OS 9 through a nearest neighbour resize-filter. Trust me, it would be bad.
Barto
PLEASE, explain to me why everything would be slightly out of place. Im thick, I just do not get it.
You need a 32x32 graphic at a certain place, you have a rendered 32x32 graphic from a vector discription, you place it at the right co-ordinate, how can it be out of place?
This is A Good Thing&trade, as you get more accurate display (if a tiny bit fuzzy). Think of the text rendering in Mac OS X currently, but for all the windows and graphics. The biggest advantage of this is the resolution independence of the system, and it will look better and better as the resolution of displays increase.
What you describe is to have co-ordinates in mm, but the boundries wrapped to the nearest pixel. That would certainly allow caching of object textures, but the edges of objects would appear closer or futher away depending where they are positioned on screen, even if the user specifies them to be the same distance (in mm). The edges of objects would also be aliased, unless you applied some kind of "smooth type" effect between pixels of the same object border.
It is certainly do-able, but there are disadvantages in the system for not much gain.
Barto
When I first learned that Apple was expected to switch to a 15.4" display on the expected new Powerbooks (up from 15.2") I was shocked to learn that this increace in screen size would actually mean a decrease in screen resolution. Rumors claimed that it would go from 1280x856 (15.2") to a resolution of 1280x800 (15.4"). This in my mind is just absurd. Dell also delivers a 15.4" laptop these days and offers a choice of two resolutions for it. The high-end has a crisp 1920x1200 WUXGA screen and the low-end offers a measly 1280x800 WXGA screen-resolution.
If the rumors are true (either for this Powerbook update of a future one) that Apple would downgrade its display on their professional line of computers so that it would match that of Dell's budget variant, I might have to think twice about making the switch I have so long yerned to make.
Personally I think that there must be middleground to find here, because while I find the current resolution of 1280x856 unsatisfactory the 1920x1200 resolution offered by Dell may, for some people, be too much. Surely Apple could offer the high-end WUXGA display as a BTO-option or use a resolution that is higher than today yet not quite as high as the one on the high-end Dell portable. Personally I hope that the last of the two options is the way forward for Apple's 15.2" (possibly 15.4") Powerbook.
What do you think?
If Apple goes with a 16:10 15.4" screen, their options would be 1280x800, 1440x900, and 1920x1200. Those are the screen resolutions we know exist at that size. In the case of a laptop, I'm generally willing to tolerate a little more dpi, because of the portability and mostly because the screen sits closer to your face most of the time, and a higher dpi is more tolerable on a crisp LCD than on a CRT display.
Apple could go with 1440x900, the same as the 17, and bump the 17 up to 1600x1024, if such a panel is available.
The availability of panels though is the key. For comfort, the dpi of the 1280 (15.2/15.4) and 1440 (17) widths is well judged.
For the 15.4" PowerBook (if it EVER gets here), I'd like to see a resolution somewhere around the 1440x900 area like Matsu suggested. Since I won't pretend like I know what the rest of the Mac buying world wants or should need like some others here, I hope it's an Apple option. More importantly, I hope that Apple uses better displays than on the 17" model which looked a bit washed out on several display models that I've seen (like the iMac). Does anyone know who the OEM on those displays are?
Originally posted by Matsu
1280x800 can be very crisp. This obsession for as Amorph put it once, "Squintronic" screens leaves me wondering. There is no practical benefit unless the UI is up to it.
At least for me, part of the problem is Aqua. Since they upped the standard font size from 9px (MacOS 1-9) to 12px, the 1152 pixel of my TiBook display roughly as much text across as did the 800x600 disply in my Wallstreet. As a coder, I need to have a lot of text on the screen simultanously.
Next is the Quartz antialiasing. It sure looks nice from a distance, but on a notebook display it is just mushy imho and the color-seams at the edges of characters bug me too. With smaller pixels, the effect should be less annoying.
] Even then, the plethora of badly done web pages, like that squinty text garbage KLD is so proud of, will keep your nose 2 inches of the screen for a good long while.
Think different
cmd + ups the font size.
Originally posted by Smircle
Think different
cmd + ups the font size.
HAha, doesn't work for those fixed ratio flash pages that some sites insist on using. Yes flash might be cool for a few effects, but it should be limited to ONLY certain effects and presentations and otherwise be part of a good old HTML page. At the end of the day legibility is more important. What's sad is a bunch of web designers and their clients think these pages look better and try to cram a bunch of small text into them, because they'd rather have that page than have it as a comliment to a nice legible HTML.