[Digikam-devel] [Bug 132047] New: Faster display of images and/or prefetch wished for

[Brad Templeton]

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
         
http://bugs.kde.org/show_bug.cgi?id=132047         
           Summary: Faster display of images and/or prefetch wished for
           Product: digikam
           Version: unspecified
          Platform: Debian testing
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: NOR
         Component: general
        AssignedTo: digikam-devel kde org
        ReportedBy: kdebug mail 4brad com


Version:           1:0.9-beta1-2 (using KDE KDE 3.5.4)
Installed from:    Debian testing/unstable Packages

You can never have enough speed.  I've been moving to digikam from other applications, and some of them are a great deal faster at display of today's large jpegs.   Some of them, like kuickshow, are just plain fast at rendering a jpeg.   Go steal its code!  pho is also fast, so is xzgv.

Another technique that can be very useful is to pre-fetch the next and/or previous image and render them in memory, or even in hidden windows/video card buffers.  RAM (screen sized ram anyway) is cheap.   Then the switch can be instant if the user has paused.  Do it in a niced background thread of course.

Digikam is much more full featured than most other programs, but once you get used to speed, it's very hard to give up speed, the features must be very compelling.  Now that digikam has the ability to do ratings, tags and comments in the large picture view (which is a must for me) the speed of going to the next picture in this view becomes important.

Some programs are also clever, and if they see the screen is much smaller than the image (as is typical) they will do a much faster cheap jpeg decode which does not extract all pixels.  Of course that means more time if the image needs zoom.   Clever programs do the mini render and have another background thread do the full render or any more complex processing.