[sci.virtual-worlds] We Need A New Language: Part 2 of 3

thinman@uunet.UU.NET (01/30/91)

       We need a new graphics programming language for Virtual
       Reality.  It should be a user-pull effort, not a vendor-push
       effort.  In this article I outline the alternatives with
       which we will be faced if we don't move now.

       Sources of fear: PHIGS

       There is a strong push by the X vendor community to
       standardize around PHIGS.  This is an old architecture and
       (apparently) meets most of the needs of the high-end CAD
       market, but doesn't serve other 3D graphics users (i.e.
       scientific visualization, virtual reality). PHIGS is a data
       list representation standard only.  It does not address
       putting a complete program into the graphics hardware; it
       assumes instead a sufficiently fast linkage between hardware
       and application for pleasant interaction speeds.

       Sources of fear: PEX

       The X Window System will soon be updated to include PHIGS
       support (PHIGS Extensions for X).  Like everything in X, PEX
       is a clumsy poorly-architected kludge.  However, it follows
       the above profile and it will succeed.  X already has
       support for animation and stereo viewing, and if there is
       enough market interest the big money behind X will pay to
       have real-time support kludged into PEX.

       Sources of fear: Renderman

       Renderman is Pixar's photorealistic rendering system
       designed around the Reyes rendering algorithm. They wrote it
       for their cinema special effects work.  Renderman is a very
       nice system for drawing static pictures.  With today's Hot
       Chips, it can be real-time simply by ignoring 99% of its
       graphics control parameters.  But, it is proprietary, and
       will stay that way.  Also, it is merely a graphics library
       and thus provides only some of what we need.

       Sources of fear: GL

       Silicon Graphic's GL is the proprietary control system for
       SGI's custom graphics platforms.  It is oriented around
       hardware-based polygonal rendering.  I don't know how
       portable it is, but its installed base puts it in the
       running.  Software-based 3D is simple enough that someone
       will write a GL emulator to achieve software portability for
       their project.  An Australian graphics package called 'VOGLE'
       claims to be something like this.

       The above systems are the top contenders in the race for a
       clumsy and unpleasant Virtual Reality software development
       environment.

Lance Norskog
thinman@cup.portal.com