My dilemma -
I have a universal app that holds a lot of images and icon (about 1,000).
In many cases i use the #2x version as is for the iPad version, in other cases it is to big even for the iPad.
I see 2 options:
Add a unique images for the ipad (large the app size).
Scale the images every time i use them?
Would like to hear notes or ideas.
Thanks
Shani
I think you will here different suggestions here. It really depends on your requirements and whether it is more important (mandatory?) for you to reduce the size of your app or make it more responsive at run-time.
Are you going to scale the images down one-by-one (meaning that in a view you will have only an image)? Or are you going to scale down a whole lot of images to display, e.g., in a table view? This will impact greatly the user experience...
As to the "every time", you could think about using a cache, so that you only resize each image just once. This would be like an intermediate solution, but again, it depends on your app (if very many images will be loaded, the cache will become very large, so in this case it could be suggested to go for option 1).
Also, the iPad 2 is pretty fast; iPad 1 a little less, so you could also take into account that.
In general, I don't think that scaling down will add much overhead, but YMMV...
Related
As I say in the title, I'm developing an iPhone app. I use nib files, I don't use any storyboard, and I know that for iPad I'll need to replace some of the controls I currently use for iPhone, since, for instance, in iPad is more suitable to use popovers in some places, and some other considerations. But I'm not sure if I'd need to create a separate nib file targeted for iPad per each nib file I have now for iPhone, or it should be just the convenient thing but not needed, or I can keep just one nib file if views are for example scroll views or table views, and just resizing things would be enough...
What I want is some guidelines to avoid redundant files and work when creating an iPad version of an existing iPhone version, and what the best practices are, since I don´t find how to handle this, programmatically speaking, neither in Apple's docs nor in posts...
Thanks in advance
EDIT. A question about dealing with icons and images: let's say I have an image view that is 50x50 in iPhone. I have two .png images for the iPhone version of this image: 50x50 and 100x100 for retina display. Let's say I need this image to be 80x80 in iPad. What should be the best way to deal with this: having 4 versions of the image (50x50, 100x100, 80x80, 160x160)? or just having the greatest versions (the 80x80 and 160x160 for iPad), and just resizing them to be smaller for iPhone? In general, what is the best practice about this, having one image file per each size you need, or just having the greater you need and fitting it to smaller sizes?
At xib files you should use the AutoLayout with constraints. That was introduced in ios6.
The key and the war with patience and time it will be setting up correctly those constraints.
One xib can be used to iPhone and iPad too, not needed 2 separate files in this way. As speed of development the 2 file are faster to develop, at least for me...
You can usually get away with not re-implementing a lot of .nib files to be iPad specific and just reuse your pre-existing ones. I have a lot of projects that do this.
That said... you usually have to reimplement the top level container .nib to be iPad specific and you really need to think about where you can take advantage of the larger screen size on iPad and adapt or create new .nibs as you see fit. While you're in any .nib... consider updating it to use AutoLayout if you can!!!
Using child view controllers (UIViewController has had childViewController since iOS 5.0) and other libs like for example SGBDrillDownController (you can find it in CocoaPods.org) might be things to take a look at.
On your images question... If the images scale well and don't get too anti-aliased looking when you go from largest to smallest rendition then going that route can make your life a lot easier. I have found however that to get a decent looking image it is best to use scalable vector art and then create a separate rendition for each size needed. Getting the best possible graphics possible is certainly a key to success.
I am developing an iPhone app with skiing destinations being listed in an UITableView. The table view can contain up to 2000 cells. Every cell has a small weather icon of 25x25px (Retina 50x50). I am already using weather icons on the detail view of every destination. These icons are twice the size: 50x50px on the 3GS and 100x100px with Retina display.
As rendering pngs in a UITableView is pretty memory consuming I am not sure if I should use the larger images also for the table view or if I should design images half the size for the table view only. Of course this would increase the download size of the app by about 300 KB.
It would be nice if anybody could give me an advice if it is best to recycle larger images in the app or if it is best to design images for every resolution needed.
Thanks in advance
Martin
You definitely want to use 2 images. If you use 1 image, you are using more memory on non-retina screens, but the same on retina screens. Using 2 images uses the least memory. There should be no problem with memory whatsoever with what you are doing, so I'd venture to say that you are keeping the images in memory after they are needed, which is unrelated to the table view code.
A good thing about UITableView is that it's memory usage is not particularly related to the cell's sizes. The little hoops you have to jump through as the datasource are there so that the table can recycle cells, keeping the memory relatively fixed with respect to memory usage of a given cell.
Recycling cells trades speed for space, so there's a more legitimate concern about scroll performance. This is related to the setup effort per cell (since they are being recycled and re-setup all the time). Image size can be factor here, but I don't think so at the scales you're considering (50^2 or 100^2). This is the kind of thing best investigated by running on real hardware.
On binary size, I think you're correct that a single version of the images would reduce the app download. It's probably not a make or break factor at 300k, but kudos for considering it.
Currently iOS supports three different art sizes: art for the 480x320 original iPhone screen, art for the hi-res 960x640 iPhone 4 screen, and art for the 1024x768 iPad screen, which in my experience is usually not the same as for the hi-res screen because of the different demands placed upon the different aspect ratio of the screen.
In a worst case, you could include three sizes for all of your artwork, and that's exactly what I did for the main element (cards) of my last game. However, it's ultimately wasteful in both download time and my time in creating them all.
What would y'all suggest as the best way to deal with the different requirements of the different iDevices? Here's a few general possibilities that I'm considering:
Include artwork in all 3 sizes.
Include artwork in just 1 size (the big one), but resize it on the fly as UIImages are created.
Include artwork in just 1 size (the big one), but make resized copies of it the first time a user starts up your app.
There are of course variants involving using some techniques for just some sizes (e.g., share artwork between the two big sizes, slightly resizing as needed). I'm interested in which method you'd use, with notes on the pros and cons of doing so. I expect the biggest cons are going to be: file size and any lag that might be caused in display or resizing of the objects. One of my biggest unknowns is how good of a job iOS does if you let it do the resizing, both in output quality and in timing.
I'm currently using artwork in all 3 sizes. If your artwork is in vector format and designed for both the iPhone and iPad aspect ratios, it's a simple batch action to create all 3 at the same time.
If you just include the highest resolution, the memory usage might be a problem with older iOS devices which have less memory to work with. The image quality and speed of resizing is not an issue, since both are excellent, if you let the OS resize the images. For me, I don't want to have additional code to handle resizing the same image for different devices. Otherwise it sounds like having just the large artwork and making sure it's resized correctly will be fine.
According to Apple, the iPhone 4 has a new and better screen resolution:
3.5-inch (diagonal) widescreen Multi-Touch display
960-by-640-pixel resolution at 326 ppi
This little detail affects our apps in a heavy way. Most of the demo apps on the net have one thing in common: They position views in the believe that the screen has a fixed size of 320 x 480 pixels. So what most (if not all) developers do is: they designed everything in such a way, that a touchable area is (for example) 50 x 50 pixels big. Just enough to tap it. Things have been positioned relative to the upper left, to reach a specific position on screen - let's say the center, or somewhere at the bottom.
When we develop high-resolution apps, they probably won't work on older devices. And if they do, they would suffer a lot from 4-times the size of any image, having to scale them down in memory.
According to Supporting High-Resolution Screens In Views, from the Apple docs:
On devices with high-resolution screens, the imageNamed:,
imageWithContentsOfFile:, and
initWithContentsOfFile: methods
automatically looks for a version of
the requested image with the #2x
modifier in its name. It if finds one,
it loads that image instead. If you do
not provide a high-resolution version
of a given image, the image object
still loads a standard-resolution
image (if one exists) and scales it
during drawing.
When it loads an image, a UIImage object automatically sets the size and
scale properties to appropriate values
based on the suffix of the image file.
For standard resolution images, it
sets the scale property to 1.0 and
sets the size of the image to the
image’s pixel dimensions. For images
with the #2x suffix in the filename,
it sets the scale property to 2.0 and
halves the width and height values to
compensate for the scale factor. These
halved values correlate correctly to
the point-based dimensions you need to
use in the logical coordinate space to
render the image.
This is purely speculation, but if the resolution really is 960 x 640 - that's exactly twice as high a resolution as the current version. It would be trivially simple for the iPhone to check the apps build target and detect a legacy version of the app and simply scale it by 2. You'd never notice the difference.
Engadget's reporting of the keynote included the following transcript from Steve Jobs
...It makes it so your apps run
automatically on this, but it renders
your text and controls in the higher
resolution. Your apps look even
better, but if you do a little bit of
work, then they will look stunning. So
we suggest that you do that
So I infer from that, if you use existing APIs your app will get scaled up. If you take advantage of new iOS4 APIs, you can get all groovy with the new pixels.
It sounds like the display will be ok but I'm concerned about the logic in my game. Will touchesBegan positions return points in the new resolution? The screen bounds will be different, these types of things could potentially be problems for me.
Scaling to a double resolution for display purpose is straight forward, but will this scalling apply to all api's that input/output a screen coordinate? If not things are going to break aren't they?
Fair enough if it's been handled extensively throughout the framework.. I would imagine there are a lot of potential api's this effects.
For people who are coming to this thread looking for a solution to a mobile web interface, check out this post on the Webkit blog: http://webkit.org/blog/55/high-dpi-web-sites/
It seems that Webkit has solved this problem four years ago.
Yes it is true.
According to WWDC it appears that apple has build it some form of automatic conversion so that the resolution for applications will not be completely off. Think up-convert for dvd to HDTV's.
My guess would be that apple knows what most of the standards developers have been using and will already be using these for an immediate conversion. Of course if you are programming an application to take advantage of the new resolution it will look much nicer than whatever the result of apples auto-conversion is.
All of your labels and system buttons will be at 326dpi but your images will still be pixel doubled until you add the hi-res resources. I am currently updating my apps. If you build and run on the iPhone 4 sim then it is presented at 50%, go to Window > Scale > 100% to see the real difference! Labels are smooth, my images look shocking!
Does anyone know what will happen with existing apps when they run on the iPhone 4.0 in terms of the new screen resolution? I am assuming, just like developing for the iPad that there should be no hard coded screen resolutions in your code.
I'd also like advice on the best way of writing robust code to work well on any device. For instance, detecting the screen resolution is not enough - on the iPad the screen is physically bigger so you can display more items on it. On the new iPhone the screen is the same physical size but higher resolution, so the likely thing is that you wont want to display more items, just higher resolution versions of them.
Any help would be useful,
Regards
Dave
EDIT: I have read the other similar posts, I guess what I really would like to know is what is the recommended way to write code for all App Store devices in a robust way so they a) all work b) make best use of the device.
UIKit has be redesigned so that old apps just work unmodified in the iPhone 4. There are then several things you can do, some programmatic and some by just adding higher resolution images to you app bundle.
Firstly, and most simply you can include new double res images that are used by your app with a suffix of #2x in the name. i.e., Event.png as well as Event#2x.png. [UIImage ImageNamed:] will automatically look for the a file with this suffix if it is running at the higher res.
All the other UIKit stuff now uses points instead of pixels. So for both old and new apps full screen is still 320 x 480 points. This pretty much means everything will work including touches, etc. Although they may now return fractions of a point.
The only real gotcha so far seems to be if you use CGBitmapContextCreate as this uses pixels and requires some jiggery - pokery.