Quality grading in DAM



  • Avatar
    Denis Kotsee

    There's no info on the exact features that are going to be available yet but there's definitely going to be some sort o rating system in place.

    Comment actions Permalink
  • Avatar
    Sherwood Botsford

    Under the hood, various ratings can be implemented with keywords.  However you often can't see all the keywords without mousing over or clicking on an image.

    You can still use them for searching.  So if you wanted to implement a quality and composition search, you would create keywords Q0 through Q6, and C0 through C6.  Then you can do searches for Keyword = Q4|Q5|Q6 AND Keyword= C5|C6


    This is clunky.


    In Aperture you have a rating system that runs 0-5 and reject.  You assign ratings with the number keys 0-5 and 9 for reject.  In addition you have flags. Either on or off.

    In use I use 0 to mean I've not looked at it.  1 means that I will probably reject it later.  2 means it is good enough to use as an illustration, (e.g. product picture) but has no artistic appeal.  3 are the backup frames.  4 is the best shot in a sequence of similar shots.  (E.g. I shoot 5 compositions, with bracketed exposures of a spruce bud covered in frost -- ONE gets a 4, four get 3's and the wrong exposures (10 of them) get 2's or rejects.  5 is reserved for pictures that are "Wow"  I only give a handful of 5's a year.


    The flagging system is a quick way to select candidate images for a particular project if they are scattered in multiple folders.  Flags are ephemeral.  It fits into my work flow like this:

    Select all pics that have keyword = Muckle Plum; rating 3 or more.
    Scan them flagging the ones that suit my purpose.

    Now look at flagged images.  Now I'm unclicking flags, and they vanish from the collection.  Try to winnow it down to 6 or so.  These 6 I will play with trying to get the look I want.

    On occasion I have wished could have more than one set of flags, but that's rare.

    In J. Anderson's case he wants a dual rating system. I've not run into a system that had that.  If presented with this as a problem, I would use keywords for the relevance issue.  A picture shot with one intent can often be used in a very different context.  Relevance is awfully slippery.  Better to have classes of keywords that denote different classes of relevance.

    E.g. have a set of mood keywords (M_Bright M_Cheerful M_Dramatic M_Sad M_Spooky M_Scary M_Cold)

    I have a big set of keywords I use for relevant time of day and weather.  E.g. ToD > Golden Hour, ToD > Sunset, ToD > Twilight, ToD > Noon Flat Light  Weather > Bright Sun; Weather > Hazy Bright; Weather > Overcast; ...  Sky > Bright Blue, Sky > Dramatic


    When you use keywords like this, you are creating Facets.  Not all images will have all facets.  An interior shot will clearly not have a Sky facet.  But facets are a way to help get the relevant keywords on each image.  


    Comment actions Permalink

Please sign in to leave a comment.