In iOS preparing graphics is simple. There are either a normal image (height x width) or a retina image which is @2x (2 times height x 2 times width).
However, since I'm new to Android, I see a ton of drawable-* folders in Eclipse where the * can be "hdpi" or "ldpi" or "mdpi" or "xhdpi" or "xxhdpi". Can someone very clearly and simply list for me what I must do to satisfy each of the display possibilities so my images will look right in each instance? I'm envisioning an answer will be a bullet list with each "*" listed and a sub-bullet list including the things that must be done.
I'd also really enjoy an answer that would start with the highest density and greatest dimension image and working down since I'll be creating in Photoshop and will be reducing quality from a master image. Thanks in advance!
In Android Studio just go to
File -> New -> Image Asset
and create your images right out of the IDE.There is an online tool for that Android Asset Studio And also there is File|New|Android Icon Set in Eclipse
i got this off of this site a while back, it still comes in handy
On Android we usually handle image sizes in units of "dp" or "dip" which stands for device independent pixel. 1 dip = 1 pixel, on a mdpi screen. There are loads of devices out there with different screen densities, not just normal and retina, so there are multiple DPI buckets a device's screen may fall into:
Note that these are buckets, so a device with a 170 dpi screen will count as an mdpi device.
Let's say that you have a vector based image in PS and you need to create an image resource for Android and you'd like to support all these screen densities. Let's say that image needs to be 100x100 dip large. So you create a 100x100 pixel version for mdpi, a 150x150 pixel version for hdpi, 200x200 for xhdpi, and 75x75 for ldpi. You can think of "mdpi - xhdpi" on Android as "normal - retina" on iOS.
As for the larges image size that you can use, I really can't say. There's no hard limit as far as I know, but the device obviously won't be able to load a 20000x20000 bitmap into memory without downsampling because of heap limits.