Re: www.humanmarkup.org

["Brandon J. Van Every" <vanevery@indiegamedesign.com>]

whitncom@lynx.eaze.net wrote:

> Has anyone looked at http://www.humanmarkup.org ?
> I'm not sure if its relevant to Interaction and Fiction, but maybe someone
> else has some ideas?
>
>  
I glanced at it.  I'm a 3D graphics guy with a limited knowledge of 
motion capture, both in terms of theoretical principles and industrial 
practice.  The range of human behaviors that one could 'mark up' is 
incredibly vast.  For instance, think of all the ways you could scrunch 
up your face.  Also there are zillions of body types that such behaviors 
can be applied to, and the problem of scaling behaviors to fit different 
body types is decidedly non-trivial.  Thus, to be technically feasible, 
a "Human Markup Language" is inevitably some kind of very limited 
subsampling of what human beings can do.  I haven't read the spec, but 
I'd expect it to contain something about a "reference avatar" that 
performs a lot of common movements that embody e-commerce problems.  For 
instance, looking in particular directions, moving the lips according to 
some phonemes in a not-too-accurate way, waving, smiling, nodding, 
pointing, etc.  I would expect a Human Markup Language to be adequate 
for putting a human face (or body) upon a UI.  In other words, I'd 
expect it to handle something like 
http://www.oddcast.com/sitepal/?promotionId=1385&bannerId=134

I would not expect such a standard to encompass anything close to 
dramatic subtlety.  First off, the data and streaming requirements are 
too heavy for XML.  Second off, in industrial practice I don't think any 
of the "high quality motion" stuff is that mature yet.  Last I checked, 
people were still getting bad input streams from their mocap that they 
had to clean up tediously by hand.  I think I last looked 2 years ago; 
there may have been some progress since then.  On the other hand, I just 
looked at whether speech synthesizers are capable of dramatic inflection 
and authoring, for purposes of film or game work.  They're not.  That's 
"next generation" stuff and it may be 5..10 years before it's delivered 
in any industrially meaningful sense.  One guy at the Audio Engineering 
Society laughed that this kind of problem has always been "5 years out" 
forever.  Which fits with my observation that for the vast majority of 
businesses, anything that's 5+ years in the future is in the infinite 
future, and nowhere near to getting done.  Most companies simply don't 
plan farther than 5 years in advance.  They wait-and-see whether someone 
else is going to make progress on something, and take some of the 
initial lumps.


Cheers,                     www.indiegamedesign.com
Brandon Van Every           Seattle, WA

When no one else sells courage, supply and demand take hold.