CamOn: A Real-Time Autonomous Camera Control System
Paolo Burelli and Arnav Jhala
Center For Computer Games Research
IT University Of Copenhagen
pabu@itu.dk, arjh@itu.dk
Abstract
This demonstration presents CamOn, an autonomous cam-
era control system for real-time 3D games. CamOn employs
multiple Artificial Potential Fields (APFs), a robot motion
planning technique, to control both the location and orienta-
tion of the camera. Scene geometry from the 3D environment
contributes to the potential field that is used to determine po-
sition and movement of the camera. Composition constraints
for the camera are modelled as potential fields for controlling
the view target of the camera.
CamOn combines the compositional benefits of constraint-
based camera systems, and improves on real-time motion
planning of the camera. Moreover, the recasting of camera
constraints into potential fields is visually more accessible to
game designers and has the potential to be implemented as a
plug-in to 3D level design and editing tools currently avail-
able with games.
Introduction
In interactive 3D applications, camera control is a key aspect
of interaction between the user and the virtual environment.
Real-time autonomous camera control has been an area of
research for several years in the graphics as well as the AI
community. A comprehensive survey of techniques devel-
oped so far in solving this problem is given by Christie et.
al. (2006).
We describe CamOn, a system that simulates the camera
as a particle moving in an Artificial Potential Fields(APFs)
to perform animation and frame composition tasks. APFs
is an technique commonly used in robotics for controlling
navigation of robots in dynamic environments; it has been
previously applied to camera control (Beckhaus, Ritter, and
Strothotte 2000) to address camera path planning for auto-
mated exploration in virtual environments. Our system ex-
tends this work by integrating frame composition constraints
to both camera positioning and orientation.
CamOn
CamOn is an autonomous camera control system capable
of generating smooth camera animations and solving frame
composition tasks. The camera is iteratively moved to con-
verge to an optimal configuration, at each iteration it takes
Copyright c© 2009, Association for the Advancement of Artificial
Intelligence (www.aaai.org). All rights reserved.
the current camera position, view direction, frame descrip-
tion and scene description as input, and returns a new camera
position as output.
Frame description requires identification of the frame sub-
jects, definition of subject importance and definition of com-
position rules; composition defines disposition of visual el-
ements in an image (Arijon 1991). Inspired by the model
proposed by Bares et. al. (2000) we translated three compo-
sition rules into constraints:
• Visibility (defines what fraction of the subject must be vis-
ible in the frame)
• ProjectionSize (defines the size of subject in the frame)
• View Angle (defines the angle from which the camera
should shoot the subject).
Artificial Potential Fields
CamOn system employs Artificial Potential Fields(Khatib
1986) to move camera and solve frame composition tasks.
We model the camera and its view target as two particles
moving in the same 3D environment. Frame composition
constraints are translated into APFs which. The APFs gener-
ated this way, along with APFs created from 3D scene geom-
etry of the environment, are used to influence the movement
of both the view point and the camera position particles.
Obstacle avoidance is modelled using repulsive forces and
frame composition is obtained by translating frame con-
straints into forces affecting both the position and the look-
at point of the camera. Each frame constraint produces an
attracting or repulsing force for camera position and an at-
tracting or repulsing force for camera view point. The sys-
tem treats these two aspect of the camera as two different
particles moving into two different potential fields.
Recasting Framing Constraints as APFs
Recasting frame constraints into APFs requires the identifi-
cation of position and orientation goals for each of the frame
constraint values. Ideal camera positions and orientations
for each constraint are modelled as low potential zones; any
other part of the solution space has a potential proportional
(the exact relation can vary from constraint to constraint) to
the distance from the ideal position and to the constraint sat-
isfaction of the corresponding camera configuration. Figure
1 shows an example of the two potential fields created by a

Figure 1: Example APF produced by a visibility constraint
on a sphere (from left to right: sphere, position APF, orien-
tation APF)
Visibility frame constraint on a sphere. Note that the distri-
bution of forces (force footprint) could be custom designed
for each object by a designer. This footprint also defines
curves along which the camera moves when no other inter-
fering forces are present in the 3D environment.
Implementation
CamOn is currently implemented as an Ogre3D 1 extension.
The library contains functions that are used to automatically
compute the motion of the camera according to the currently
loaded camera profile. A camera profile is the description of
camera’s visual constraints on frame composition and it is
defined as a set of subjects (each with an importance value)
and a set of viewing constraints applied to each subject.
CamOn translates the camera profile into two potential
fields, one affecting camera position and the other one cam-
era orientation, each of these is a linear combination of
scalar weighted potential fields, where each potential field
corresponds to a frame constraint and each scalar value to
the relative subject importance. The system animates two
particles along the two potential fields until they both reach
a stable configuration, camera position and camera look-at
point both follow the movement of the two particles.
Demonstration Overview
The demonstrative application shows a non-interactive in-
tro sequence showing the main character (a green ninja) and
the enemies he has to avoid (4 steady robots and 2 walk-
ing robots). Users can watch the default camera animation
(a predefined sequence of camera profiles), can release the
camera from CamOn control, and move it manually or can
interactively decide to shoot any character in the scene by se-
lecting a specific camera profile. The camera definition file
included in the demo can also be modified to test the camera
animations corresponding to different constraint values and
weights. Figure 2 shows sample camera shots for the given
profile(Fig. 2(a)). In the first shot, shown in Figure 2(b)
the robots are walking in the same direction and the cam-
era maintains the shot to face the front of the two robots. In
Figure 2(c) the robots are facing the opposite directions and
the camera adapts to find a shot that maximizes constraint
satisfaction.
1Object-oriented Graphics Rendering Engine -
http://www.ogre3d.org
(a) Camera profile definition
(b) Robots facing same di-
rection
(c) Robots facing opposite
directions
Figure 2: Camera showing the solution of a three-quarter
front shot of two robots in two different conditions
In this demonstration, the user can interact with the sys-
tem in the following ways:
• Change camera profile values and switch between camera
profiles in real-time.
• Interact with the camera during a non-interactive session
and jump back to the non-interactive walk-through at any
time and from any location in the world.
Conclusions
In this demonstration, we present CamOn - a real-time au-
tonomous camera control system - that is based on multiple
APFs. The system solves camera motion and frame com-
position problems by modelling frame constraints and scene
geometry as multiple APFs. The demonstrative application
allows the user to modify camera profiles, observe the re-
sult of changing camera parameters in real-time and interact
with the camera during an animation by changing the cam-
era location and orientation by hand. The camera system
converges to the desired location and orientation from any
initial location in the 3D world.
References
Arijon, D. 1991. Grammar of the Film Language. Silman-
James Press LA.
Bares, W.; McDermott, S.; Boudreaux, C.; and Thainimit,
S. 2000. Virtual 3d camera composition from frame con-
straints. In MULTIMEDIA ’00, 177–186. ACM.
Beckhaus, S.; Ritter, F.; and Strothotte, T. 2000. Cubi-
calpath - dynamic potential fields for guided exploration in
virtual environments. In PG ’00. IEEE Computer Society.
Christie, M., and Olivier, P. 2006. Camera Control in Com-
puter Graphics. 89–113. Eurographics Association.
Khatib, O. 1986. Real-time obstacle avoidance for manip-
ulators and mobile robots. Int. J. Rob. Res. 5(1):90–98.