Newsgroups: comp.graphics,sci.psychology,sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech!howland.reston.ans.net!news.sprintlink.net!mv!mv.mv.com!rmorris
From: rmorris@mv.mv.com (Robert Morris)
Subject: Perceptual Cues in Direct Vol. Rendering Env.
Message-ID: <D29t6I.How@mv.mv.com>
Nntp-Posting-Host: mv.mv.com
Sender: usenet@mv.mv.com (System Administrator)
Organization: MV Communications, Inc.
Date: Thu, 12 Jan 1995 02:20:41 GMT
Lines: 111
Xref: glinda.oz.cs.cmu.edu comp.graphics:68137 sci.psychology:33271 sci.image.processing:11866


Here's a question.

   In traditional graphics environments (with polygonal representation of
objects), certain perceptual cues are provided that allow the user to
determine the spatial location and orientation of surfaces.  These
include contours, shading and texture (i.e., texture gradient in
perspective projections).  Local or global illumination models - the
same idea holds.

   The relative simplicity of this type of environment gives theses
perceptual cues a strong ability to convey spatial information
to the user.

   In Direct Volume Rendering environments, the situation is more complex.
When illumination falls on a surface, it it transmitted in a relatively
complex way to interior subsurfaces.  Some light is reflected back,
transmitted to interior regions, scattered to interior regions and
moves about in a generally more complex way.  One surface casts a
shadow on another partially opaque surface, which may cast another
more complex shading pattern on another surface, and on and on. . .

   Both types of models can be made more complex by adding the full-range
of ways to simulate the interaction of light with matter . . . refraction,
diffraction, reflection, scattering. . . but the idea is that Direct
Volume Rendering is more complex by nature.

   Now take into account that Human visual perception capabilities evolved
in the presence of light and its' interaction with matter.  Our ancestors
tended to live and thrive in proportion to the accuracy of their visual
perception of our physical world.  Natural selection drove the evolution
of our eye-brain visual apparatus and, as a result, the apparatus is
based upon the foundation of a physically accurate model of light.  To the
degree that a lighting simulator departs from a physically accurate
model of light, the resulting images may fail to take advantage of natural
human perception capabilities.

   Classical rendering techniques are similiar to much of what is seen
in the real world.  Both deal predominately with the interaction of light
with matter at or near the surface of objects.  Direct volume rendering
largely departs from this type of light-matter interaction.  In nature,
one rarely gets to observe a 3D region of space with multiple, complex
interior regions that overlap and are semi-transparent.  When such material
does exist, human survival does not depend on the accurate perception
of such a region of space.

   I say this:  "The human visual system has not evolved to accurately
                 interpret many of the images that are routinely generated
                 using Direct Volume Rendering".

   Real-time interaction and processing out of homogeneous regions of space
help get around this.  A number of other simplifying tricks probably
exist that help mitigate this also - I won't think of them or note them
here.  Just becomming familiar and practiced with this type of imagery
also helps perception.

   I'm considering static imagery only.

   When you get to many transparent, overlapping and competing surfaces, at
different intensities, colors, texture, spatial orientation, etc - the
visual system quickly fails to perceive the spatial situation embodied
in the volume.  Using the traditional graphics techniques, the situation
is less complex and the human visual system has been better equipped
by evolution to handle the task.

   Please note that all of the above is not necessarily factual.  I can't
show that the statements about the evolution of the human perceptual
system is true, for instance.

   Before this gets long-winded let me ask the question.

   "Has anybody seen perceptual psychology experiements that elaborate
   on the effectiveness of the traditional perceptual cues in a
   DIRECT VOLUME RENDERING environment.  In addition, a discussion
   of new types of cues that are specific to DVR environments might
   also be of use (if there are any)?"

   Any reference that supports or rejects my idea that the human visual
   system works best with a physically accurate model of light (because
   it evolved using the real thing) would also be of great help.

   Does anybody know of any perceptual tricks that would fool our
   eye-brain system into perceiving data features in a direct volume
   rendering environment.  I don't mean image/volume processing
   techniques, but ways to manipulate the perceptual cues to use
   them to better advantage in this environment - or to use them
   to better advantage when we want to see specific types of data
   features.

I hope a lot of this made sense to you.  Any references or ideas will
be deeply appreciated.

Thanks!!!

Rob Morris




   






-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Robert O. Morris    Institute for Visualization and Perception Research (IVPR)
                    UMASS/Lowell  Email: rmorris@cs.uml.edu
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
