Apple Secretly Made Interactive iBooks Author For "SuperHD" Retina Display iPad 3
Ladies and Gentlemen, I present to you iBooks on iPad 3. You will never look at a normal "computer screen" the same way. For the first time ever, digital books will be indistinguishable from printed material. I never thought I would live to see this day.
[...]
Original images generated natively from iBooks Author:
"No one can tell you what the Matrix is, you have to see it for yourself"
Accordingly, the SuperHD Retina iBook2 screenshot above can not be fully appreciated on regular monitors. But if you imagine those images in a 10" screen, then, you'll get a brief feel of what it is going to be like on the iPad 3. But in any case, seeing, feeling and touching "books" on an iPad 3 will be unprecedented. This is big. This is the next iGoldRush?.
Thanks for the post. This does get me excited.
Note the text above I bolded. There will plenty that will say magazines and other printed material has more DPI than the iPad and that iPad might be pretty decent but it's not nearly as good as printed material. Please try not to beat them with your iPad 3... I'd hate for you damage a perfectly good iPad.
Note that the screens above are generated natively from the currently-shipping iBooks Author. For several years now, Keynote has supported "SuperHD" resolutions, allowing for some digital projections of these material to be done by seamlessly combining 1080p projectors, such as for the product launch below:
Now imagine most of the Macs above replaced by an iPad 3.
Also, now that digital cinema has evolved significantly, all you would need is an iPad 3 and 4K digital projector (iPad 3 resolution sits somewhere between "2K" and "4" resolutions.
Given the Keynote/iWork software similarity to iBooks Author, Retina Display iBooks are now clearly already available in iBooks Author.
Note that the screens above are generated natively from the currently-shipping iBooks Author. For several years now, Keynote has supported "SuperHD" resolutions, allowing for some digital projections of these material to be done by seamlessly combining 1080p projectors, such as for the product launch below:
Now imagine most of the Macs above replaced by an iPad 3.
Also, now that digital cinema has evolved significantly, all you would need is an iPad 3 and 4K digital projector (iPad 3 resolution sits somewhere between "2K" and "4" resolutions.
I'm sorry, but as a consumer, you had me at 'Note'.
I will hereby call this technology 'KFAT' for 'Kick Fucking Ass Technology', and look forward to seeing soon.
I'm sorry, but as a consumer, you had me at 'Note'... I will hereby call this technology 'KFAT' for 'Kick Fucking Ass Technology', and look forward to seeing soon.
... Last bit of R&D from me, gotta go have lunch.
Keynote '09 on Mac is Apple iPad 3 "SuperHD" Retina Display Ready
The GPU does not need to be 4x the power, or 4x the "VRAM", because of 4x the pixels.
Because by that token, for 2D or 3D games, going from 1024x768 to 1920x1080 on a PC means you need a GPU that is 2.64 times as powerful, with 2.64 times the "VRAM". Which is certainly not the case.
The 4x only directly applies to the final framebuffer, whereby you need 12MB for the output as opposed to 3MB.
Everything else is dependent on the implementation. GPU power and "VRAM" requirements do not scale linearly with pixels pushed.
Your wrong. It scales with the number of pixels you need to render.
(Note that I never said anything about memory use being 4 times as much.)
And that's ofcourse a worst case scenario.
But as any 3D game developer knows that's wat counts.
I still don't see it. Lets say I have a 1024x768 display and I want to display the color black over the entirety of it. You're saying that all things being equal I need 4x the power to render that same image on a 2048x1536 display? That sounds inefficient to me.
The point is that you need 4x the rendering speed for the most demanding tasks not for every rendering. But if you don't have the 4x rendering speed your game will drop frame rates (and stutter) at some points in the game, while the game with the 4 times lower resolution runs fine.
So the 4x faster is a worst case requirement and essential to have.
If the iPad3 ~must~ have 4x the performance, they would have to have 8 cores for the iPad3, or use the Rogue GPU which would have to be shown to do 4x the performance of the ~dualcore~ SGX543.
When the iPad3 comes out, if the GPU doesn't have 4x the performance of a dualcore SGX543MP4, that is, as per Wikipedia, 14.4*4 = 57.6 GFlops, I buy you a beer? If not, you buy me a beer?
Remember a GeForce 8500 GT does only 43 GFlops in singlecore. The PS3 is estimated to be around this ballpark, though graphically it is probably not as good as a 8500GT at the 1280x720 resolutions it runs at before upscaling. If the iPad 3 could do 1024x768 graphics at 8500 GT resolutions, with 1080p upscaling to HDTV, then I will be indescribably happy.
If the iPad3 ~must~ have 4x the performance, they would have to have 8 cores for the iPad3, or use the Rogue GPU which would have to be shown to do 4x the performance of the ~dualcore~ SGX543.
When the iPad3 comes out, if the GPU doesn't have 4x the performance of a dualcore SGX543MP4, that is, as per Wikipedia, 14.4*4 = 57.6 GFlops, I buy you a beer? If not, you buy me a beer?
If the numbers are correct (and I think they are) logic dictates that that is the case.
It seems a lot of graphics power, but the device would be underpowered if it wasn't that fast.
It could be ofcourse that the GFlops are close to the 57.6 GFlops needed (say 50 or so) that would be acceptable I think.
My estimate was that they had to choose the Rogue version to be able to achieve the performance needed, and this version is available.
It could also be that Apple has made an adaptation of the current version of the GPU with, say 400MHz quad cores.
We will see.
(Nice of you to show the numbers, that definitely improves the discussion.)
Remember a GeForce 8500 GT does only 43 GFlops in singlecore. The PS3 is estimated to be around this ballpark, though graphically it is probably not as good as a 8500GT at the 1280x720 resolutions it runs at before upscaling
The RSX in the PS3 can do either 240 or 400 GFLOPS (Wikipedia says around 400, other people say around 240 GFLOPS). On top of that, the PS3's CPU can do ~150 GFLOPS (6 SPUs available for games).
However the GPU architecture in PowerVR is more efficient due to the tile based renderer, but not enough to make up for the above in raw power.
People are saying it needs 4x to match if they boost the resolution like the rumors all say. Maybe the current GPU exceeds what currently runs. If (random number use here) the system is capable of handling 1280*1024 currently before hitting slowdown, then they would only need to add 2.4x current power, not 4x. That is a big chunk cut off requirements.
I don't know how things go honestly, but it's just an example. If the current iPad 2 already outclasses its current resolution, things don't have to stretch quite as far.
People are saying it needs 4x to match if they boost the resolution like the rumors all say. Maybe the current GPU exceeds what currently runs. If (random number use here) the system is capable of handling 1280*1024 currently before hitting slowdown, then they would only need to add 2.4x current power, not 4x. That is a big chunk cut off requirements.
I don't know how things go honestly, but it's just an example. If the current iPad 2 already outclasses its current resolution, things don't have to stretch quite as far.
The RSX in the PS3 can do either 240 or 400 GFLOPS (Wikipedia says around 400, other people say around 240 GFLOPS). On top of that, the PS3's CPU can do ~150 GFLOPS (6 SPUs available for games).
However the GPU architecture in PowerVR is more efficient due to the tile based renderer, but not enough to make up for the above in raw power.
Yeah, I was a bit iffy on the GFlops for the PS3, not sure where exactly I got the low number, the other estimates seem very high but in reality the graphics performance of the PS3 is not that great. So maybe not a good comparison because GFlops is not really directly related to visual performance.
In any case though that's one metric in measuring the PowerVR chips.
Quote:
Originally Posted by jnjnjn
If the numbers are correct (and I think they are) logic dictates that that is the case.
It seems a lot of graphics power, but the device would be underpowered if it wasn't that fast.
It could be ofcourse that the GFlops are close to the 57.6 GFlops needed (say 50 or so) that would be acceptable I think.
My estimate was that they had to choose the Rogue version to be able to achieve the performance needed, and this version is available.
It could also be that Apple has made an adaptation of the current version of the GPU with, say 400MHz quad cores.
We will see.
(Nice of you to show the numbers, that definitely improves the discussion.)
J.
Cheers, so is our bet on or off? $5 Paypal on the table. Deal or no deal? GFlops or any other official metric.
Quote:
Originally Posted by SolipsismX
@ sunilraman,
Wouldn't that be 2K video on the iPad 3 if you are doing a pixel for pixel rendering?
I put it as "3K" in quotes because it's sort of inbetween 2K and 3K resolution, plus it's 4:3.
2K video is 2048×1080 and 4K is 4096×2160. Red's "3K" Scarlet has a 3072x1620 or so sensor.
Retina Display on iPad 3 is QXGA [2048x1536], based on the evidence. The only thing is no video is made in QXGA resolution or ratio.
In my mockup above it's cropped and downscaled from a Red 4K sample to QXGA ratio.
Suffice to say if there is ever a chance of iPad 2/3/4/5 Retina Display capable of displaying QXGA video, the best results would be using a 3K or better digital camera or scanned film to have really nice video on iPad Retina. As we know, even 2K video will have to be upscaled on an iPad Retina, let alone BluRays.
Also, as shown above, for digital projections of superHD presentations if the iPad Retina can video mirror QXGA, you would need the 4K digital projector which is what I think digital cinema has standardised on (2K is too low, and there's no common 3K digital cinema projectors).
Very exciting times. The video resolution is simply insane on an iPad Retina. I forsee digital film really taking on the iPad Retina as a portable, simple monitor of sorts if it's ever able to interface with the iPad Retina. Certainly for digital still cameras no doubt the iPad will be essential for mobile viewing of high-res digital stills ~ as some have mentioned.
And if the iPad Retina can play 1080p video, you can see it in native pixels on the iPad Retina. No mean feat on 10".
I imagine as we speak China is churning out portable Blu Ray players with this kind of 10" panel technology.
Quote:
Originally Posted by SSquirrel
If the current iPad 2 already outclasses its current resolution, things don't have to stretch quite as far.
Not so sure about that, even the latest Infinity Blade on iPad 2 doesn't have a freely moving user-controlled camera, so I think the iPad 2 is mostly maxxed out as it is.
Your wrong. It scales with the number of pixels you need to render.
(Note that I never said anything about memory use being 4 times as much.)
And that's ofcourse a worst case scenario.
But as any 3D game developer knows that's wat counts.
J.
Dear (jn)+,
The assertion made in your posts is that the processing required to support high resolutions is proportional to number of pixels. The evidence provided is that this is would be logical. You may or may not be aware that much of the underlying foundation for *3D* programming involves linear algebra applied to polygons. Matrix transformations can be used to determine position in a 3D space, probably to apply lighting effects, and other effects. As a result, polygons can be a constraining factor and performance specifications for 3D cards are often expressed in terms of number of polygons, more so than "pixel count". If this is wrong and your understanding of implementation and hardware requirements is more detailed, then please add to this discussion. If not, it would be much appreciated if you could refrain from saying "four times more pixels requires four times more power, of course" which may be more of a guess than carefully considered.
The assertion made in your posts is that the processing required to support high resolutions is proportional to number of pixels. The evidence provided is that this is would be logical. You may or may not be aware that much of the underlying foundation for *3D* programming involves linear algebra applied to polygons. Matrix transformations can be used to determine position in a 3D space, probably to apply lighting effects, and other effects. As a result, polygons can be a constraining factor and performance specifications for 3D cards are often expressed in terms of number of polygons, more so than "pixel count". If this is wrong and your understanding of implementation and hardware requirements is more detailed, then please add to this discussion. If not, it would be much appreciated if you could refrain from saying "four times more pixels requires four times more power, of course" which may be more of a guess than carefully considered.
Not so sure about that, even the latest Infinity Blade on iPad 2 doesn't have a freely moving user-controlled camera, so I think the iPad 2 is mostly maxxed out as it is.
I'm not sure if it is an on rails game due to worries about overdoing it against the gpu or if it is b/c they had a hard time coming up w/a good way to have the look around functionality work. Probably some of both, but I don't know
Comments
This is massive.
[Indulge me one cross-post]
Apple Secretly Made Interactive iBooks Author For "SuperHD" Retina Display iPad 3
Ladies and Gentlemen, I present to you iBooks on iPad 3. You will never look at a normal "computer screen" the same way. For the first time ever, digital books will be indistinguishable from printed material. I never thought I would live to see this day.
[...]
Original images generated natively from iBooks Author:
iPad 2
http://dl.dropbox.com/u/21386792/low_res_01.png
http://dl.dropbox.com/u/21386792/low_res_02.png
iPad 3
http://dl.dropbox.com/u/21386792/high_res_01.png
http://dl.dropbox.com/u/21386792/high_res_02.png
"No one can tell you what the Matrix is, you have to see it for yourself"
Accordingly, the SuperHD Retina iBook2 screenshot above can not be fully appreciated on regular monitors. But if you imagine those images in a 10" screen, then, you'll get a brief feel of what it is going to be like on the iPad 3. But in any case, seeing, feeling and touching "books" on an iPad 3 will be unprecedented. This is big. This is the next iGoldRush?.
Thanks for the post. This does get me excited.
Note the text above I bolded. There will plenty that will say magazines and other printed material has more DPI than the iPad and that iPad might be pretty decent but it's not nearly as good as printed material. Please try not to beat them with your iPad 3... I'd hate for you damage a perfectly good iPad.
Now imagine most of the Macs above replaced by an iPad 3.
Also, now that digital cinema has evolved significantly, all you would need is an iPad 3 and 4K digital projector (iPad 3 resolution sits somewhere between "2K" and "4" resolutions.
Given the Keynote/iWork software similarity to iBooks Author, Retina Display iBooks are now clearly already available in iBooks Author.
Note that the screens above are generated natively from the currently-shipping iBooks Author. For several years now, Keynote has supported "SuperHD" resolutions, allowing for some digital projections of these material to be done by seamlessly combining 1080p projectors, such as for the product launch below:
Now imagine most of the Macs above replaced by an iPad 3.
Also, now that digital cinema has evolved significantly, all you would need is an iPad 3 and 4K digital projector (iPad 3 resolution sits somewhere between "2K" and "4" resolutions.
I'm sorry, but as a consumer, you had me at 'Note'.
I will hereby call this technology 'KFAT' for 'Kick Fucking Ass Technology', and look forward to seeing soon.
I'm sorry, but as a consumer, you had me at 'Note'... I will hereby call this technology 'KFAT' for 'Kick Fucking Ass Technology', and look forward to seeing soon.
Keynote '09 on Mac is Apple iPad 3 "SuperHD" Retina Display Ready
iPad 2 presentations ("100% crop")
iPad 3 presentations ("100% crop")
Generated natively from Keynote '09 on Mac.
Full Images:
http://dl.dropbox.com/u/21386792/hig...ote_01.001.png
http://dl.dropbox.com/u/21386792/hig...ote_01.002.png
Note that the first slide has super-crisp text and vector elements.
In the second slide, we are able to display two 720p videos on one slide.
The GPU does not need to be 4x the power, or 4x the "VRAM", because of 4x the pixels.
Because by that token, for 2D or 3D games, going from 1024x768 to 1920x1080 on a PC means you need a GPU that is 2.64 times as powerful, with 2.64 times the "VRAM". Which is certainly not the case.
The 4x only directly applies to the final framebuffer, whereby you need 12MB for the output as opposed to 3MB.
Everything else is dependent on the implementation. GPU power and "VRAM" requirements do not scale linearly with pixels pushed.
Your wrong. It scales with the number of pixels you need to render.
(Note that I never said anything about memory use being 4 times as much.)
And that's ofcourse a worst case scenario.
But as any 3D game developer knows that's wat counts.
J.
I still don't see it. Lets say I have a 1024x768 display and I want to display the color black over the entirety of it. You're saying that all things being equal I need 4x the power to render that same image on a 2048x1536 display? That sounds inefficient to me.
The point is that you need 4x the rendering speed for the most demanding tasks not for every rendering. But if you don't have the 4x rendering speed your game will drop frame rates (and stutter) at some points in the game, while the game with the 4 times lower resolution runs fine.
So the 4x faster is a worst case requirement and essential to have.
J.
If the iPad3 ~must~ have 4x the performance, they would have to have 8 cores for the iPad3, or use the Rogue GPU which would have to be shown to do 4x the performance of the ~dualcore~ SGX543.
When the iPad3 comes out, if the GPU doesn't have 4x the performance of a dualcore SGX543MP4, that is, as per Wikipedia, 14.4*4 = 57.6 GFlops, I buy you a beer? If not, you buy me a beer?
Remember a GeForce 8500 GT does only 43 GFlops in singlecore. The PS3 is estimated to be around this ballpark, though graphically it is probably not as good as a 8500GT at the 1280x720 resolutions it runs at before upscaling. If the iPad 3 could do 1024x768 graphics at 8500 GT resolutions, with 1080p upscaling to HDTV, then I will be indescribably happy.
OK, final one for now:
http://www.youtube.com/watch?v=2Ukmp...eature=related
Wouldn't that be 2K video on the iPad 3 if you are doing a pixel for pixel rendering?
If the iPad3 ~must~ have 4x the performance, they would have to have 8 cores for the iPad3, or use the Rogue GPU which would have to be shown to do 4x the performance of the ~dualcore~ SGX543.
When the iPad3 comes out, if the GPU doesn't have 4x the performance of a dualcore SGX543MP4, that is, as per Wikipedia, 14.4*4 = 57.6 GFlops, I buy you a beer? If not, you buy me a beer?
If the numbers are correct (and I think they are) logic dictates that that is the case.
It seems a lot of graphics power, but the device would be underpowered if it wasn't that fast.
It could be ofcourse that the GFlops are close to the 57.6 GFlops needed (say 50 or so) that would be acceptable I think.
My estimate was that they had to choose the Rogue version to be able to achieve the performance needed, and this version is available.
It could also be that Apple has made an adaptation of the current version of the GPU with, say 400MHz quad cores.
We will see.
(Nice of you to show the numbers, that definitely improves the discussion.)
J.
Remember a GeForce 8500 GT does only 43 GFlops in singlecore. The PS3 is estimated to be around this ballpark, though graphically it is probably not as good as a 8500GT at the 1280x720 resolutions it runs at before upscaling
The RSX in the PS3 can do either 240 or 400 GFLOPS (Wikipedia says around 400, other people say around 240 GFLOPS). On top of that, the PS3's CPU can do ~150 GFLOPS (6 SPUs available for games).
However the GPU architecture in PowerVR is more efficient due to the tile based renderer, but not enough to make up for the above in raw power.
I don't know how things go honestly, but it's just an example. If the current iPad 2 already outclasses its current resolution, things don't have to stretch quite as far.
People are saying it needs 4x to match if they boost the resolution like the rumors all say. Maybe the current GPU exceeds what currently runs. If (random number use here) the system is capable of handling 1280*1024 currently before hitting slowdown, then they would only need to add 2.4x current power, not 4x. That is a big chunk cut off requirements.
I don't know how things go honestly, but it's just an example. If the current iPad 2 already outclasses its current resolution, things don't have to stretch quite as far.
Infinity Blade (II).
J.
The RSX in the PS3 can do either 240 or 400 GFLOPS (Wikipedia says around 400, other people say around 240 GFLOPS). On top of that, the PS3's CPU can do ~150 GFLOPS (6 SPUs available for games).
However the GPU architecture in PowerVR is more efficient due to the tile based renderer, but not enough to make up for the above in raw power.
Yeah, I was a bit iffy on the GFlops for the PS3, not sure where exactly I got the low number, the other estimates seem very high but in reality the graphics performance of the PS3 is not that great. So maybe not a good comparison because GFlops is not really directly related to visual performance.
In any case though that's one metric in measuring the PowerVR chips.
If the numbers are correct (and I think they are) logic dictates that that is the case.
It seems a lot of graphics power, but the device would be underpowered if it wasn't that fast.
It could be ofcourse that the GFlops are close to the 57.6 GFlops needed (say 50 or so) that would be acceptable I think.
My estimate was that they had to choose the Rogue version to be able to achieve the performance needed, and this version is available.
It could also be that Apple has made an adaptation of the current version of the GPU with, say 400MHz quad cores.
We will see.
(Nice of you to show the numbers, that definitely improves the discussion.)
J.
Cheers, so is our bet on or off? $5 Paypal on the table. Deal or no deal? GFlops or any other official metric.
@ sunilraman,
Wouldn't that be 2K video on the iPad 3 if you are doing a pixel for pixel rendering?
I put it as "3K" in quotes because it's sort of inbetween 2K and 3K resolution, plus it's 4:3.
2K video is 2048×1080 and 4K is 4096×2160. Red's "3K" Scarlet has a 3072x1620 or so sensor.
Retina Display on iPad 3 is QXGA [2048x1536], based on the evidence. The only thing is no video is made in QXGA resolution or ratio.
In my mockup above it's cropped and downscaled from a Red 4K sample to QXGA ratio.
Suffice to say if there is ever a chance of iPad 2/3/4/5 Retina Display capable of displaying QXGA video, the best results would be using a 3K or better digital camera or scanned film to have really nice video on iPad Retina. As we know, even 2K video will have to be upscaled on an iPad Retina, let alone BluRays.
Also, as shown above, for digital projections of superHD presentations if the iPad Retina can video mirror QXGA, you would need the 4K digital projector which is what I think digital cinema has standardised on (2K is too low, and there's no common 3K digital cinema projectors).
Very exciting times. The video resolution is simply insane on an iPad Retina. I forsee digital film really taking on the iPad Retina as a portable, simple monitor of sorts if it's ever able to interface with the iPad Retina. Certainly for digital still cameras no doubt the iPad will be essential for mobile viewing of high-res digital stills ~ as some have mentioned.
And if the iPad Retina can play 1080p video, you can see it in native pixels on the iPad Retina. No mean feat on 10".
I imagine as we speak China is churning out portable Blu Ray players with this kind of 10" panel technology.
If the current iPad 2 already outclasses its current resolution, things don't have to stretch quite as far.
Not so sure about that, even the latest Infinity Blade on iPad 2 doesn't have a freely moving user-controlled camera, so I think the iPad 2 is mostly maxxed out as it is.
Your wrong. It scales with the number of pixels you need to render.
(Note that I never said anything about memory use being 4 times as much.)
And that's ofcourse a worst case scenario.
But as any 3D game developer knows that's wat counts.
J.
Dear (jn)+,
The assertion made in your posts is that the processing required to support high resolutions is proportional to number of pixels. The evidence provided is that this is would be logical. You may or may not be aware that much of the underlying foundation for *3D* programming involves linear algebra applied to polygons. Matrix transformations can be used to determine position in a 3D space, probably to apply lighting effects, and other effects. As a result, polygons can be a constraining factor and performance specifications for 3D cards are often expressed in terms of number of polygons, more so than "pixel count". If this is wrong and your understanding of implementation and hardware requirements is more detailed, then please add to this discussion. If not, it would be much appreciated if you could refrain from saying "four times more pixels requires four times more power, of course" which may be more of a guess than carefully considered.
Dear (jn)+,
The assertion made in your posts is that the processing required to support high resolutions is proportional to number of pixels. The evidence provided is that this is would be logical. You may or may not be aware that much of the underlying foundation for *3D* programming involves linear algebra applied to polygons. Matrix transformations can be used to determine position in a 3D space, probably to apply lighting effects, and other effects. As a result, polygons can be a constraining factor and performance specifications for 3D cards are often expressed in terms of number of polygons, more so than "pixel count". If this is wrong and your understanding of implementation and hardware requirements is more detailed, then please add to this discussion. If not, it would be much appreciated if you could refrain from saying "four times more pixels requires four times more power, of course" which may be more of a guess than carefully considered.
Shhhh... I'm trying to get a free beer from him!
Not so sure about that, even the latest Infinity Blade on iPad 2 doesn't have a freely moving user-controlled camera, so I think the iPad 2 is mostly maxxed out as it is.
I'm not sure if it is an on rails game due to worries about overdoing it against the gpu or if it is b/c they had a hard time coming up w/a good way to have the look around functionality work. Probably some of both, but I don't know