The increasing performance of PCs
over recent years has made the use of computer-generated video a practical
proposition for visualisation without being prohibitively expensive.
The images on the left are thumbnail stills from videos produced by
Envision. Click on them to see an enlargement of the image and to see a clip
from the video.
Animation is a logical extension of visualisation. Instead of a single animated image,
a sequence is generated, generally with the viewpoint moving along a flightpath
through the scene depicted. As there are generally 25 frames per second of
animation (the same as European TV standards), many hundreds and often
thousands of images are produced. Each image generally takes a few tens of
seconds to be rendered on the computer, sometimes leading to days of computer
time to complete the job.
Once all the individual frames have been generated, they are fitted together
into video clips on the computer and from that point on, can be edited and
captioned just like any other video material.
As its name suggest, videomontage is the motion-picture counterpart to
photomontage. In the same way that a
photomontage consists of a base photograph with a computer rendered still image
superimposed on it, a videomontage consists of a base video with an animated
image superimposed on it.
The base video is shot on site using a video camera. The camera moves used
at that time naturally define the camera moves that will be seen in the
finished montage. Great care is therefore needed to design the moves so that
they will eventually look sensible and show the things they are intended to
show, even though the subject matter of the video is wholly absent at the time
The whole process of videomontage is dependent for its success on the
accuracy with which the imaginary camera of the computer graphics world mimics
the movement of the actual camera used on site. This process, called
match-move, is now an established trade in its own right in the motion picture
As match-move is very time-consuming (and therefore very expensive),
Envision has also developed an alternative approach to videomontage. Instead of
using live footage shot on site as a base image, we have produced some
videomontages with panoramas spliced together out of still photographs shot on
site as a background. These panoramas can be made to behave as if they were
cylindrical or spherical shells around the virtual camera in such a way that
the perspective is correct in the finished video. The virtual camera cannot
move of course, but it can pan and zoom at will and the precise combination of
pans and zooms can be refined through many iterations instead of having to be
achieved once on site with a video camera.
While not strictly video work, Envision has made considerable use of desktop
virtual reality through the medium of VRML (Virtual Reality Modelling
Language). VRML worlds are 3D computer rendered representations of a scene with
the ability to walk or fly around and through them in real time using VRML
viewing software. VRML has proved effective in a number of situations where
clients have wished to explore the visual effects of a proposed development
over a large area, particularly where serial vision from roads is important.
Envision has also made use of VRML as a means of rendering animations. We
developed our own software to record a flightpath from a session using a VRML
viewer and subsequently to reproduce that path as a sequence of frames,
grabbing each in turn and saving it to file.