Loading ...
Sorry, an error occurred while loading the content.

1549RE: [humanmarkup] food for thought

Expand Messages
  • Bullard, Claude L (Len)
    Aug 24, 2001
    • 0 Attachment
      HyTime grew out of music work and there have been
      different efforts to use SGML and musical notation
      systems. MIDI being essentially control information
      to a synthethizer for patches, GM instruments,
      expression parameters, etc. could certainly be
      XMLized. The contribution of that to HumanML
      seems scarce though, so I'm sure where that goes.

      In the genre example, I borrowed <par> and <seq>
      from SMIL 1.0 to demonstrate nesting of chronemic
      information to describe actions. It is one
      of those overlaps in that scheduling
      is vital to almost any simulation or analysis
      of time-based information. If you examine
      the interpolation components of VRML, you
      see similar concepts for key frame-based animation.

      A problem of HumanML will be to decide just
      what is high level information and what should
      be added in by transformation given that the
      downtranslation target will have it's own
      representations the closer one gets to the
      implementation objects. EMOTE has very
      interesting concepts for shaping the motions
      of an animated character using a level that
      seems to be between what HumanML has described
      and what H-anim might use. I am unsure if this
      is middleware data or goes in HumanML or H-anim. I am
      inclined to think it is a middleware dataset
      in which HumanML uses a simple intensity value
      to scale an emotion, an EMOTE engine target
      gets that and creates a more detailed LMA
      based dataset, then the final rendering in
      say H-anim or SVG is created. Hard to say
      and other opinions on that would be appreciated.

      My concern is mostly that HumanML stay high level
      and very easy to use so that the derived languages
      get most of the work and the HumanML is repurposable.
      Even as we look at domains, we have to ask ourselves
      first what the role of HumanML is quite precisely
      in the abstraction layer.


      Ekam sat.h, Vipraah bahudhaa vadanti.
      Daamyata. Datta. Dayadhvam.h

      -----Original Message-----
      From: Kurt Cagle [mailto:kurt@...]

      This is perhaps off topic, but is anyone aware of a MIDI to XML schema map?
      I have to wonder about the intersection between an XML-ized MIDI, SMIL and
      HumanML ...
    • Show all 8 messages in this topic