[livesupport-dev] playlists and queries
  • This is a multipart message in MIME format.
    --=_alternative 003DFB95C1256EC4_=
    Content-Type: text/plain; charset="iso-8859-1"
    Content-Transfer-Encoding: quoted-printable

    Hi,

    I read an interesting post about where we could possibly go with playlists =

    and queries as opposed to playlists and static files being served off the=20
    file system.

    It's an interesting idea, because in effect we will be making our own=20
    queries through the archive module, which will then pass the file from the =

    file system. But what this guy seems to be arguing is that you can store=20
    the queries themselves, and then create possibly random sequences based on =

    rules you create. An interesting idea, at least, via the Unmediated=20
    weblog. - doug

    http://dream.sims.berkeley.edu/~ryanshaw/wordpress/2004/06/28/playlists-que=
    ries-rules/

    Playlists, Queries, Rules
    Filed under: Media Automation Information Organization & Retrieval ? ryan @=
    8:58 pm=20
    Today I spent some time discussing playlists with Lucas Gonze, which got me=
    thinking about the spectrum with simple playlists on one=20
    end and full-blown cinematography on the other. Both essentially describe=20
    sequences of media selected and arranged in a purposeful manner.
    What Lucas made me realize is that a playlist needn't be a static record:=20
    it can be a query, a suggestion or template that can be realized by a=20
    number of possible content sequences. For example, a playlist entry that=20
    specifies only ["I Love You" by Cole Porter] could be satisfied by either=20
    Ella Fitzgerald's or Frank Sinatra's version of the tune (or countless=20
    others).=20
    This realization brought to mind something from a paper I read recently:
    The rules used to generate establishing shots are based on cinematographic =

    principles to maintain continuity? [For example, we have a rule that]=20
    defines an establishing shot as a combination of two sequences of which=20
    both shots are in color, the focus of the first has a wider angle then the =

    next shot, and the weather conditions are similar. Note that the rule can=20
    also be described as a query that returns all shot combinations that match =

    a certain description: we found that in this (and probably many more)=20
    applications the difference between a rule and query is rather artificial.
    This makes the thread of continuity along the spectrum a little clearer.=20
    Initially we have static playlists, which arrange specific pieces of=20
    content in a sequence. Then "smart" playlists, which define a space of poss=
    ible sequences as a query over a media=20
    database. More sophisticated queries return not just individual pieces of=20
    content that match the query but combinations or subsequences that match=20
    certain criteria. These more sophisticated queries can be viewed as rules=20
    which formalize the techniques involved in creating time-based media. In=20
    film or video these are the rules of cinematography such as those that=20
    govern continuity editing. In the realm of music these are the rules follow=
    ed by DJs for selecting and mixing sets.
    Looked at this way, the recent mainstreaming of the playlist phenomenon=20
    may be the first step on the road to a future of widespread media=20
    meta-production.=20
    --=_alternative 003DFB95C1256EC4_=
    Content-Type: text/html; charset="iso-8859-1"
    Content-Transfer-Encoding: quoted-printable



    Hi,



    I read an interesting post about whe=
    re we could possibly go with playlists and queries as opposed to playlists =
    and static files being served off the file system.




    It's an interesting idea, because in=
    effect we will be making our own queries through the archive module, which=
    will then pass the file from the file system. But what this guy seems to b=
    e arguing is that you can store the queries themselves, and then create pos=
    sibly random sequences based on rules you create. An interesting idea, at l=
    east, via the Unmediated weblog. - doug




    http://dream.sims.berkeley.edu/~ryan=
    shaw/wordpress/2004/06/28/playlists-queries-rules/




    /28/playlists-queries-rules/"> Roman">Playlists, Queries, Rules

    Filed under: http://dream.sims.berkeley.edu/%7Eryanshaw/wordpress/category/media-automat=
    ion/">Media Automat=
    ion
    =3D"http://dream.sims.berkeley.edu/%7Eryanshaw/wordpress/category/informati=
    on-organization-retrieval/"> oman">Information Organization & Retrieval
    =3D3 face=3D"Times New Roman"> — ryan @ 8:58 pm

    Today I spent some time discussi=
    ng playlists with
    color=3Dblue face=3D"Times New Roman">Lucas Gonze ze=3D3 face=3D"Times New Roman">, which got me thinking about the spectrum =
    with simple playlists on one end and full-blown cinematography on the other=
    . Both essentially describe sequences of media selected and arranged in a p=
    urposeful manner.

    What Lucas made me realize is th=
    at a playlist needn't be a static record: it can be a query, a suggestion o=
    r template that can be realized by a number of possible content sequences. =
    For example, a playlist entry that specifies only ["I Love You" by Col=
    e Porter] could be satisfied by either Ella Fitzgerald's or Frank Sinatra's=
    version of the tune (or countless others).

    This realization brought to mind=
    something from a
    ?abstractnr=3D1469"> >paper I read recent=
    ly:

    The rules used to generate estab=
    lishing shots are based on cinematographic principles to maintain continuit=
    y… [For example, we have a rule that] defines an establishing shot as=
    a combination of two sequences of which both shots are in color, the focus=
    of the first has a wider angle then the next shot, and the weather conditi=
    ons are similar. Note that the rule can also be described as a query that r=
    eturns all shot combinations that match a certain description: we found tha=
    t in this (and probably many more) applications the difference between a ru=
    le and query is rather artificial.

    This makes the thread of continu=
    ity along the spectrum a little clearer. Initially we have static playlists=
    , which arrange specific pieces of content in a sequence. Then
    ef=3Dhttp://www.smartplaylists.com/> es New Roman">"smart" playlists es New Roman">, which define a space of possible sequences as a query over =
    a media database. More sophisticated queries return not just individual pie=
    ces of content that match the query but combinations or subsequences that m=
    atch certain criteria. These more sophisticated queries can be viewed as ru=
    les which formalize the techniques involved in creating time-based media. I=
    n film or video these are the rules of cinematography such as those that go=
    vern
    <=
    font size=3D3 color=3Dblue face=3D"Times New Roman">continuity editing u>
    . In the realm of musi=
    c these are
    xing.html>the rules=
    followed by DJs for selecting and mixing sets
    face=3D"Times New Roman">.

    Looked at this way, the recent m=
    ainstreaming of the playlist phenomenon may be the first step on the road t=
    o a future of widespread media meta-production.

    --=_alternative 003DFB95C1256EC4_=--

    ------------------------------------------
    Posted to Phorum via PhorumMail