Trying f-spot for (almost) the first time

[Benjamin Kahn <xkahn ximian com>]

I've just started trying f-spot out as my photo management application.
I checked out the sources from CVS yesterday (October 21st, 2004) and
compiled.  Here are some of my thoughts and comments.

When adopting a photo application, most users probably already have a
good collection of images.  So the first step is to import them all into
the new program.  In my case, I have them stored on a server.  So I
needed to copy them into F-Spot.

I saw no way inside of F-Spot to do this automatically, so I SCP'd to
the files from the server to the local machine.  But, the connection was
pretty slow (25k/s!) and the amount of data fairly large (okay, only a
Gig of data...) so I got impatient while loading the images and imported
the directory multiple times.  This lead to duplicate images.

So feature requests from this:  

      * Have a way to import images from online albums similar to
        export.  A plugin system for different types of online
        repositories would be great, but supporting directories of files
        and Gallery would be a great start.
      * During import, if the image is already in the catalog, don't
        import it again.  (Plus, what happens if I import an image
        twice, select one instance and pick Edit -> Delete from Drive?)
      * I should be able to select a directory as a source for images.
        Images added to this directory should automatically import.
      * Lots of context information was lost in the import.  I kept the
        images organized in folders which would have made GREAT tag
        names.  For example, 2003/Jan/14 should have created three tags:
        2003, Jan, and 14.  All the images in that directory should have
        been given those tags. 
      * The thumbnails displayed come from the thumbnails embedded in
        the exif data which is defined to be 160x120 always.  While this
        makes importing really fast, it means that large icons always
        display badly (until a real thumbnail is generated) and rotated
        images are displayed wrong (again, until a real thumbnail is
        generated).  Either there should be a background process for
        creating real thumbnails, or the correct thumbnails should be
        produced at import time.

Bug list:

      * While importing the incomplete list of images, the new images
        sometimes wouldn't show up in the list, or would appear blank.
      * When scrolling quickly through the list, thumbnails often don't
        appear.  When this happens, this message arrives on the console:
        BUG: Filing another request of order 26 (previously 25) for
        `/home/xkahn/.thumbnails/large/cfef809df6a3aabc6b6b79fdd654a31c.png'
      * Importing a directory was very confusing somehow.  The file
        dialog made it very difficult since it kept OPENING the
        directory I was trying to import.  For some reason it took me a
        while to notice that the "Open" button was enabled even when
        nothing was selected.  Also, once a directory was selected, it
        was hard to unselect it.  (You have to hold down ctrl and click
        that directory again.)

Next I wanted to play around with the interface a little to get a feel
with what I can do with F-Spot.  I looked at tags...

Feature Requests:

      * When you assign a tag that has no associated image yet (say...
        the Other tag) the tag's icon becomes that first image.  This
        was pretty surprising at first, and I can't remove the image
        completely.  (Okay, okay...  Why would you want a tag that
        doesn't have an icon?  How will you be able to tell when that
        tag is applied?)  Maybe this is a bug report asking that all top
        level tags come with icons already.  Or maybe a tag without an
        icon should be represented by a text label?
      * The checkboxes in the left hand sidebar are somewhat confusing
        to me.  I selected a number of pictures and clicked the checkbox
        for the Other tag.  Oops!  The checkbox is intended as a
        filtering system, and all the images disappeared.  The checkbox
        should be a tag assignment system.  (Yes, I know about the
        right-click menu, dragging tags to images and dragging images to
        tags, and the Tags -> Attach Tag to Selection and Tags -> Remove
        Tag from Selection.  These methods seem less obvious and
        confusing.)  Filtering should be a different mode from
        assigning.
      * Tags should be auto-created and assigned at import time as much
        as possible.

Bugs:

      * Tags listed in the left sidebar should all be uniformly sized.
        There shouldn't be tags that are very tall (because of their
        icon) and those that are very short. (because of their lack of
        icon)
      * When an icon is assigned to a tag, the tag spacing isn't updated
        to hold the image, and the image is chopped off.
      * When assigning a non-top level tag to an image, the top level
        tag icon is displayed as being associated.  The actual tag icon
        should be shown.
      * The Tag menu doesn't have an accelerator.
      * The menuitem "Remove Tag from Selection" doesn't have an
        accelerator.

Next, I wanted to do some light editing on the images I had imported.
First up is rotation.  I tend to always turn the camera the same way
when taking pictures, so I selected 4 images right away that all needed
to be rotated left.  On a hunch, I right clicked one of the images and
saw the rotate menu item and selected it.  

Bugs:

      * Because of the thumbnail import bug listed above, rotating
        images can be very frustrating.  (Images appear to be rotated
        twice since the thumbnail is wrong at first.)

Now I wanted to examine a number of images to make sure it wasn't
blurry, find which one was best, etc.  Double clicking on an image works
to view it, as does selecting the right tab.  Once viewing the image,
things degrade a little.

Feature Requests:

      * Standard GNOME zoom keys should work:  Ctrl-+ (Zoom In) and
        Ctrl-- (Zoom Out) and Ctrl-= (Full Size) and Ctrl-0 (Fit to
        Screen)
      * Clicking and Dragging across the image creates a crop box.  It
        should pan the image instead.  (Cropping is important I guess,
        but it's more likely that I'm examining an image, esp if the
        image is larger than the screen.  Cropping, or any edits, should
        be a mode.)  
      * If I unzoom the image too far, I end up back at the list of
        images.  Yeah, my camera does that too, but it feels a little
        weird in an application.  It should just make the image smaller.

Bugs:

      * The mouse wheel looks like it's trying to zoom the image.  (The
        image becomes sharper for a moment and then smooths out again as
        the wheel turns.)  But the zoom isn't working.  Actually, it's
        weirder than that.  If I zoom the image in and try the
        mousewheel, the image pans (very slowly) in relation to the area
        of the image the mouse is hovering over.

I know, I know.  Put these problems in bugzilla!  Okay, I will.  But
wanted to jot down my notes from trying things out first.