Abstract
This paper is on the use of 3D
data for content creation in TV- and film productions.
Computer graphics methods are used for a wide range of
applications in content creation, like the planning of
productions, for pre-visualisation on set and when 3D graphics
appear in the final programme. The degree to which 3D graphics
are involved ranges from their use in the planning phase
of a production without any graphical elements in the final
programme, to films that are entirely computer generated.
For most types of programme it is in between and typically
involves a mixture of virtual and real scene content. Well
known examples are special effects in movies, which insert
virtual objects or characters into real camera footage
or virtual studios in TV- productions that insert a presenter
or actor into a completely virtual environment. Since the
application of computer graphics is still very expensive
usually only those components that can not be filmed will
be generated virtually.
The final high-quality graphics
for film- and many TV-productions are mostly generated
off-line in a post-production
phase. For on-set pre-visualisation and many TV-productions,
that are either broadcast live or have a lower budget
than feature film productions, the graphics are generated
in
real-time. The budget and the decision whether the image
generation has to be in real-time constrain applicable
techniques. For real-time TV productions usually only
simple 2D scene composition can be used. In virtual
studio systems
for example, a camera image of the presenter or actor
is taken in a (real) studio and then keyed out and
overlayed
onto a synthesised (virtual) image.
How convincing the
composited result looks in terms of realism depends
on the quality of the optical integration
of the virtual and real scene components. In a full
optical integration virtual and real objects are optically
interacting
with each other, that means they are occluding each
other and cast shadows. With a 2D composition these
optical
interactions can only be established in a very limited
way. For sophisticated
optical interactions more 3D knowledge of the real
(and virtual) scene is needed and the scene composition
is
done in the 3D domain.
The technical challenges
of those productions that integrate virtual and real
scene components
are the
subject of
this paper.
This document was originally
published in : 3D Videocommunication: Algorithms,
concepts and
real-time systems in human
centred communication,
Oliver Schreer (Editor), Peter Kauff (Editor),
Thomas Sikora (Editor)
ISBN: 0-470-02271-X, Wiley, July 2005.
|
