Resolution Independence

Posted:
in General Discussion edited January 2014
I've been reading about this on the internet and just want to make sure I have (basically) the idea right on how it effects me -



Images/text will look better when scaled. The control+mouse wheel zoom is not resolution independent, and everything becomes pixelated and bad looking - but will later become crystal clear when it is independent.



My desktop icons appear to be independent, because they look good when I scale them down to 16x16 to 128x128. They may just be large icons scaled down and their largest size is 128x128 for all I know, just using it as an example =)



So do I understand the basics of this ok?

Comments

  • Reply 1 of 4
    nvidia2008nvidia2008 Posts: 9,262member
    Quote:
    Originally Posted by Avor View Post


    I've been reading about this on the internet and just want to make sure I have (basically) the idea right on how it effects me -



    Images/text will look better when scaled. The control+mouse wheel zoom is not resolution independent, and everything becomes pixelated and bad looking - but will later become crystal clear when it is independent.



    My desktop icons appear to be independent, because they look good when I scale them down to 16x16 to 128x128. They may just be large icons scaled down and their largest size is 128x128 for all I know, just using it as an example =)



    So do I understand the basics of this ok?



    You sort of have the basics. 8) ... I am not sure how they intend to adjust control-mousewheelzoom in Leopard 10.5. Below is some of my ideas, people will have different views/suggestions.



    But you are right. Essentially, it has to do mostly with how the window graphics (borders, header, and text is drawn). One should be able to size it independently of the screen resolution. For example, if I had a scrawny ass 1920x1200 ("scrawny ass" is an official spec name okay? ) screen then if my menu bar is only 15 pixels high, that's too tiny. Or if my palettes are only 5 pixels square, that might be too tiny.



    It means that menu bars, scroll bars, window shading, window buttons, and text of dialog boxes, etc. have to be scalable to virtually any resolution. The current idea is that a "vector-based template" will be used.



    It is a bit more complex. Additionally so when it comes to icons. Your icons "appear" to be independent, because icon standards for Mac OSX specify it should look good between 16x16 all the way up to 128x128. Here's a hint though: there are different versions of the icon for different sizes. Usually three sizes, like a 128x128 version, 64x64 and a 16x16. If you play around with the sizes you'll see sometimes the small version is really a slightly different icon in fact. What is great is that OSX will resample (ie smoothly resize) the icon for intermediate sizes, like the icons in the dock, for example.



    Resolution-independent icons should essentially mean that the icon is all vector, that means, even if the icon was 1000x1000 it should not look pixelated. However, they may have to make different vector "versions" when it comes to icons. Because sometimes for tiny icons like 20x20 just taking a big icon and resampling it down to 20x20 may not look good. That's where smart icon design comes in. At low pixel counts icon designs need to be "simpler" - more clear and contrasty, less gradients, whereas bigger (50x50 and up) you want more shades, tones, curves, gradients, and 128x128 to 1280x1280 you want a version that can scale as if you were working with a nice vector illustration in Adobe Illustrator.



    Some example of how this is done is in iPhoto'06. The thumbnail scaling goes from very small to almost full-screen. In this case, the thumbnails are scaled on-the-fly but are also run off a thumbnail index ("snapshots") of some sort. In other words, I want to be able to scale from super-tiny to super-big on-the-fly and everything should be smooth and snappy, but I will need some "hints" from pre-cached (hard disk or RAM) intermediate snapshots.



    It is a mouth-watering proposition, IMO, and indeed something very very interesting to see how this plays out in Leopard 10.5 User Interface, and the "Illuminous" or whatever the new 10.5 UI is supposed to be called.



    It will clearly be a milestone in the development of user interfaces in general, and will leave Vista in the dust. In fact, UI resolution independence was the promise of Flash-based web apps - since it is all vector, your web app should look as good as it is 800x600 or 1920x1200 etc. etc. But Apple will do it in a solid fashion (solid as in excellent)...



    Particularly seeing that Leopard Mac OS X 10.5 is going to be a highly scalable operating system. For example, running on anything from iPhone, through to AppleTV, to the rumoured ultraportable 10" MacNano (?name?) to a MacPro 8core running four 30" displays. Alongside CoreImage and CoreAnimation scaled in all these situations, it's going to be hella interesting.
  • Reply 2 of 4
    nvidia2008nvidia2008 Posts: 9,262member
    The one sticking issue is web graphics. A JPEG, PNG, GIF is a bitmap, no two ways about it. Scale it up and it will look pixelated. This sucks.



    Web page zooming should work differently in a "pure resolution independent" environment. Firefox kinda does that. If you use tile-able bitmap backgrounds, or just colours, CSS, etc etc etc scaling a web page up and down should be smooth, snappy, and sharp. Challenging. For ads, for example, it would have to be Flash vector-based so that the browser would scale that Flash section appropriately.



    For the future of resolution-independent web page experiences, we will initially have to look at Cocoa applications in Mac OS X 10.5 and see how they handle window elements, UI bits and pieces, text, and vector graphics. Also how bitmap graphics are handled, for example, some high-res and low-res versions of bitmaps will need to be "cached" like in iPhoto'06 to some extent.



    In the meantime generally if I use Firefox 2.0 on Mac or PC, increasing or decreasing text size while leaving the bitmaps untouched is the best for me, and one of the closest examples we have of approaching resolution independence in web browsers.



    Technically there is scope here for new Cocoa-developed Web Browsers - since Web2.0+ and all this is supposed to be all metadata and meta-meta-data - structure independent of content. The original promises of CSS and XML etc.



    Again, we have the tech to describe all sorts of text, UI elements, and vector graphics. But a photograph, that's still pixels that can be compressed, but can't be "metadata-ised" -- though of course Google Images has made good strides in this regard, but that's analysing a bitmap and coming up with metadata, the reverse, metadata and rendering an image, is close (think 3D Games, particle effects and shaders) but photorealism and wide range of synthesised imagery off simple metadata is 10 years down the track*... In the meantime what may need to happen is that a JPEG or GIF downloaded by the browser is of a higher-resolution than the actual layout in the page -- to allow for "scaling up overhead" if the user decides to "zoom in" (make ALL the elements in the web page bigger). Flash has broken some ground in this regard, if you import a JPEG into Flash at a high-resolution, when playing back the Flash box in a web page at smaller sizes, it *kinda* scales reasonably. But it used to be an unnecessary CPU-eating process, that's why Apple is pushing resampling bitmaps to CoreImage, offloading these kind of processes to the GPU instead. iPhoto'06, AFAIK, if you have a GMA950Intel or above GPU, all the resampling in the thumbnail view is done mostly by the GPU not CPU like in Flash-scaling.



    *Anyone remember VRML? WTF happened to it? But re-visiting it in several years will not be too bad a prospect. For example, let's say I want to have a website for the all new Mustang Convertible V12-ZeroEmission-MrFusion** 2015 model. Text and vector graphics, layout, UI elements, okay, all resolution-independent in the web browser. Now what about pictures of the car itself? Or the sexy (assuming similar misogynist values [not that that's really bad, like whatevs, you know..] down the line) lady next to the car - boom! all 3D synthesised, right there photorealistic in the web browser. Voila. VRML Strikes Back !!! ... It can be specific angles that the web designers like you to see, or of course there may be a section for QTVR-styled stuff. Restricted to an extent, for example, if you wanted to look up the skirt of the model next to the car you wouldn't be allowed to see that angle. But who knows. It's 2015....



    **BackToTheFuture. For the oldskoolers here

    (for me OldSkoolers means late-20, 30-40somethings)
  • Reply 3 of 4
    nvidia2008nvidia2008 Posts: 9,262member
    In fact *DING* *LIGHT BULB* --- the future of bitmaps is 3D. No doubt about this anymore. For 2010 through to 2025, clearly capturing an "image" will move to capturing a "scene". Initially photogrammetry and other techniques as a hybrid 3D-bitmap capture model, to, approaching 2020, being able to capture an entire landscape and say people in a scene ALL IN 3D -- for later 3D-holographic-style extrapolation and so on. WHOA. MY MIND JUST BLEW. MINDGASM.



    You think Photoshop is great now? In 10-25 years it will look like a baby smearing baby food on the table. You won't be "taking pictures", you will be essentially "capturing 3D data of a scene" -- this scene then "played back" - "viewed" - in a not only an almost resolution-independent fashion, but time and space- independent fashion as well.



    For example, you and your loved one snaps yourself with the Bay Bridge in the background. When "viewing" this. You could change the lighting - say you took it in bright daylight. But actually, what if you wanted to see it in "Romantic" mode - voila, the "image" will be "rendered" with a different lighting set so that 3D (pure or initially hybrid-photogrammetry) elements can be shifted around. Another example would be, you didn't like the spot where you were standing. No problemo. You get to move where you stood to a better position, and while you're at it, completely delete the annoying tourists blocking a certain part of the background.



    You think "PhotoShopping Fraud" is bad, or tacky, wait till the "image" is a freely-manipulated full-3D scene. Just as 2D imagery can now no longer be trusted, so too will holographic/ full-3D "images" going into the next 25 years. WHOA. FRACK ME.



    http://en.wikipedia.org/wiki/Photogrammetry and http://en.wikipedia.org/wiki/Bullet_time will be considered the WrightBrothers' equivalent for "visual data capture" in this century.
  • Reply 4 of 4
    irelandireland Posts: 17,614member
    All icons in Leopard will be drawn to 521x512 pixels, so when you xoom in they will look much better, all text should be resolution independent too, so when you zoom in ti make that bigger it will look bigger but shouldn't pixel-ate. A great example is the iPhone, just look at when Steve Zooms in you can still read the text fine, that's the gist of it.
Sign In or Register to comment.