What is a realistic image? This is an age-old question in art, and a contemporary
question in computer graphics. This book provides a modern answer involving the
computer and a new definition of realism.
The classic definition of realism has been veridical realism. Does the picture pass
the comparison test? That is, would an observer judge the picture to be real? This is
traditionally described by Pliny’s story (in Book 35 of his Natural History) of the ancient
painter Zeuxis who painted a picture of a boy carrying some grapes, and when
the birds flew up to the picture, he approached the work and, in irritation, said, “I
have painted the grapes better than the boy, for if I had rendered him perfectly, the
birds would have been afraid.”
Nowadays the ultimate in fooling the eye is special effects in the movies. Almost
every movie involves hundreds of special effects that are seamlessly combined with
live action. It is impossible to tell what is real and what is synthesized. Equally amazing
are full-length, computer-generated pictures such as Shrek. Although few would
be fooled into believing these worlds are real, it is more the artistic choice of the storyteller
than a technological limitation. A major achievement in the last two decades
is that computers allowed us to achieve veridical realism of imagined scenes.
Besides direct comparison, there are other definitions of real. Masters such as
Vermeer used optical devices to aid them in painting realistic pictures, and modern
photorealists such as Richard Estes paint over a projected image of a photograph.
Thus, another definition of real is to be traced or copied from an image. In this sense
the montage of composite layers in a movie is photoreal, since different elements
come from different film sequences. There are many other definitions of realism. For
example, real can mean a choice of subject matter, such as everyday life versus a
myth or an idealized form.