- Atlas
- Computer Science
- Computer Animation
Computer Animation
Computer animation is the art and engineering of producing motion by computation: an animator describes how a character, object, or camera should change over time, and the computer generates the frames.
7 frameworks, from 1963 to today · Computer Graphics · Computer Science
Overview
Computer animation is the art and engineering of producing motion by computation: an animator describes how a character, object, or camera should change over time, and the computer generates the frames1. It sits inside computer graphics, but its central question is different from rendering's. Rendering asks how to draw a convincing picture; animation asks how to specify change over time in a way that is expressive, controllable, and affordable. That question has driven the field through several distinct eras—from hand-set keyframes, to physics, to performance capture, to learned motion—and each era changed what animators could make and what skills they needed.
Keyframes and the Interpolating Machine
The earliest computer animation inherited its vocabulary from hand-drawn animation. Traditional animators drew "key" poses and let assistants "in-between" them2. The first computational idea was to formalize this: an animator specifies keyframes—poses or values at particular times—and the computer interpolates between them. Interpolation means filling in intermediate values smoothly, usually with curves such as splines (mathematical curves that pass through or near control points)12.
Early demonstrations established the idea. In the early 1960s, Ivan Sutherland's Sketchpad showed interactive manipulation of graphical objects, and researchers at Boeing and elsewhere produced short animated sequences of moving objects and cameras. By the 1970s, Ed Catmull's work on animation of smooth surfaces and the Utah graphics community's experiments made keyframe interpolation a standard tool. The practical payoff was enormous: an animator no longer had to draw every frame. The limitation was equally clear—interpolation produces smooth, mechanical motion. It knows nothing about weight, momentum, or intent.
Procedural Animation: Rules Instead of Frames
Procedural animation answered that limitation by generating motion from rules rather than from explicit poses3. Instead of keying every step, the animator writes a procedure—a set of instructions or mathematical functions—that produces motion. A classic early example is the "particle system," introduced by Reeves in the 1980s for effects like fire, smoke, and rain: thousands of simple particles are born, move under simple rules, and die, producing complex natural-looking motion from a short program. Procedural methods also produced flocking birds and swarming insects, where each agent follows local rules and the group behavior emerges.
Procedural animation's strength is scale and naturalness in effects; its weakness is directability. A director who wants a specific splash to land at a specific moment cannot easily key a particle system. This tension—between emergent complexity and artistic control—reappears throughout the field's history.
Character Animation and Rigging
The hardest problem was the human figure. A character is not a rigid object; it is an articulated skeleton covered by skin. Rigging is the craft of building that skeleton—a hierarchy of joints and bones—and binding the skin (or surface) to it so that when joints rotate, the surface deforms plausibly4. Early rigs were simple; by the 1990s, production rigs included facial controls, corrective shapes, and layered deformation systems.
The landmark achievement was the 1995 film Toy Story, the first fully computer-animated feature, made by Pixar15. Its characters required rigs sophisticated enough to support expressive acting over ninety minutes. The film demonstrated that computer animation could carry narrative and emotion, not just spectacle. Character animation also raised the "uncanny valley" problem—the observation that near-human figures can feel more disturbing than stylized ones—which shaped how studios chose between realistic and stylized characters for years.
Physically Based Animation
Where keyframes and procedures struggled, physics offered another route: simulate the equations of motion6. Physically based animation lets the computer compute motion from forces, masses, and constraints6. Cloth, hair, water, smoke, and rigid bodies are simulated rather than posed3. The animator sets up the physical situation—gravity, wind, material stiffness—and the simulation produces the motion.
This approach gave effects unprecedented realism. The challenge was control: a simulation may not do what the director wants. The field responded with "art-directable" simulation—methods that let animators guide a simulation toward a desired outcome while keeping physical plausibility. The tension between realism and directability remains a defining feature of effects work today.
Motion Capture and Performance Capture
Motion capture offered yet another answer: record real motion and apply it to a digital character17. Early systems tracked markers on a performer's body; later systems captured facial expression and even the subtle motion of hands and eyes. Performance capture—the combination of body, face, and sometimes voice—let actors drive digital characters directly7. Landmark uses include Gollum in The Lord of the Rings films and the Na'vi in Avatar, where performance capture became a central production method.
Motion capture solved the realism problem for human motion but introduced new ones: cleaning data, retargeting motion to differently proportioned characters, and blending captured motion with keyframed or simulated motion. It also changed the animator's role—from posing every frame to directing, editing, and augmenting captured performance.
The Production Pipeline
None of these techniques stood alone. Computer animation matured as a production pipeline—an organized sequence of stages from story and modeling through rigging, animation, simulation, lighting, and rendering. Pipeline thinking is itself a historical development: studios built proprietary systems to manage complexity, share assets, and coordinate large teams. The pipeline made possible feature-length computer-animated films and increasingly complex visual effects, and it shaped how animation is taught and practiced as an industrial craft.
Machine Learning Animation
The most recent turn applies machine learning to motion. Learned methods can generate motion from examples, synthesize faces and gestures, and assist animators by predicting plausible poses or filling in motion. These approaches promise speed and accessibility, but they raise questions about control, authorship, and the relationship between data-driven motion and artistic intent. The field is still working out where learned methods fit alongside keyframes, physics, and capture.
The Present
Today's computer animation is plural. Keyframe interpolation remains the backbone of character animation; rigging and pipeline craft are still central; physics drives effects; motion capture supplies human realism; and machine learning is being folded into tools for generation and assistance. The open questions are the old ones in new form: how to give animators precise control over complex, emergent motion; how to make learned methods directable and trustworthy; and how to balance realism against expression. The history of the field is a history of expanding the animator's vocabulary—from poses, to rules, to physics, to performance, to data—without ever fully resolving the tension between what the machine can generate and what the artist intends.
References
- Computer animation (introduction, Interpolation, Film and television, Animation methods), Wikipedia
- Interpolation (computer graphics), Wikipedia
- Procedural animation, Wikipedia
- Skeletal animation, Wikipedia
- Pixar, Wikipedia
- Physically based animation, Wikipedia
- Motion capture, Wikipedia
The frameworks of Computer Animation
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1963-Present
Keyframe Animation
How do we specify poses over time and fill in between?
- 1983-Present
Procedural Animation
How can simple rules generate complex natural motion?
- 1985-Present
- 1986-Present
Physically Based Animation
How can we simulate motion from physical laws?
- 1990-Present
- 1995-Present
Production Pipeline Organization
How do we organize many people and tools to produce animation?
- 2015-Present
Machine Learning Animation
How can machines learn to generate and assist motion?
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
Procedural Animation reacted against Keyframe Animation
Procedural animation arose as an alternative to keyframe interpolation, which produced smooth but mechanical motion lacking weight and intent. Reeves's particle systems (1983) generated motion from rules rather than explicit poses, defining itself against the keyframe approach.
Carried by William T. Reeves.
Skeletal animation grew out of Keyframe Animation
Skeletal animation extended keyframe interpolation from rigid objects to articulated figures: animators keyed joint angles and the system interpolated between poses. The skeleton hierarchy and skinning were built on the same keyframe-and-interpolate machinery.
Physically Based Animation reacted against Keyframe Animation
Physically based animation emerged because keyframe interpolation could not produce realistic cloth, water, or collisions without painstaking manual posing. Simulation replaced hand-set keys with equations of motion, though keyframes remained for control.
Motion capture rival of Keyframe Animation
Motion capture offered an alternative to keyframing human motion by recording real performance. It solved the realism problem that keyframe interpolation struggled with for human figures, though captured data was later blended with keyframed motion.
Physically Based Animation influenced by Procedural Animation
Particle systems and other procedural methods for effects were absorbed into physically based animation, which simulated the underlying forces rather than using ad hoc rules. Many effects pipelines combine procedural emission with physical simulation.
Carried by William T. Reeves.
Motion capture influenced by Skeletal animation
Motion capture data is applied to skeletal rigs, so the joint hierarchy and skinning developed for skeletal animation became the target representation for captured motion. Retargeting and cleanup methods built directly on rigging techniques.
Machine Learning Animation influenced by Motion capture
Motion capture databases provided the large motion datasets used to train learned models for motion synthesis and prediction. Learned methods often aim to reproduce or generalize captured performance.
Machine Learning Animation influenced by Keyframe Animation
Machine learning animation often uses keyframe poses as training data or as control handles, and learned models are integrated into keyframe-based tools. The keyframe interface remains the way animators direct learned motion.
Beyond the field
Where Computer Animation's ideas came from in other fields, and what they led to there: the strongest 12 of 23 links.
Keyframe Animation influenced by Studio-System Character Animation in Animation Theory
The Disney studio's character animation practice, codified in the 1930s through techniques like the exposure sheet and pose-to-pose timing, supplied the model that early computer keyframe systems imitated. Ivan Sutherland's Sketchpad and later animation systems adopted the pose-and-inbetween structure that studio animators had established.
Physically Based Animation led to Position-Based Dynamics in Physical Simulation
Position-based dynamics was developed within physically based animation as a fast, stable alternative to force-based integration for interactive graphics. It grew directly out of the simulation techniques of physically based animation.
Physically Based Animation influenced Physics-Based Character Control in Physical Simulation
Physics-based character control took the articulated rigid-body simulation, contact handling and numerical integration of physically based animation and applied them to controlling simulated characters. The character controllers are built on the same simulation machinery.
Physically Based Animation influenced Art-Directable Simulation in Physical Simulation
Art-directable simulation grew out of physically based animation: the same numerical integrators, constraint solvers and collision handling developed for physically based animation were extended with artist-facing control handles and goals. The later field's makers took those simulation techniques and added directability on top.
Machine Learning Animation influenced by Physics-Based Character Control in Physical Simulation
Physics-based character control established the simulated-articulated-body setting and control objectives that learning-based animation later adopted, replacing hand-tuned controllers with learned policies. Deep reinforcement learning character controllers built directly on this simulation framework.
Machine Learning Animation influenced by Machine Learning for Physics Simulation in Computational Physics
Learned physics simulators such as neural-network fluid and cloth models supplied the simulation backbone that machine-learning animation systems use to generate motion without hand-authored physics. The same research community moved from predicting physical fields to driving animated characters and scenes.
Keyframe Animation influenced by Spline and subdivision surface modeling in Computer Graphics
Spline interpolation from computer-aided design and graphics supplied the mathematical basis for interpolating between keyframes. Early keyframe systems used spline curves to generate smooth inbetweens automatically.
Procedural Animation influenced by Procedural modeling in Computer Graphics
Procedural modeling techniques, such as those developed for generating complex geometry algorithmically, supplied the generative approach that procedural animation applied to motion. The idea of specifying rules rather than explicit data passed from modeling to animation.
Procedural Animation influenced by Procedural programming in Programming Paradigms
Procedural programming's separation of algorithmic procedure from data provided the programming model on which procedural animation was built. Animation systems expressed motion as procedures evaluated over time rather than stored frame data.
Procedural Animation influenced Art-Directable Simulation in Physical Simulation
Procedural animation supplied the idea of generating motion from rules and parameters rather than keyframes, which art-directable simulation adopted for its high-level control layer. The later work combined procedural control with physical simulation so artists could steer simulated characters.
Skeletal animation influenced by Studio-System Character Animation in Animation Theory
Studio character animation's hierarchical pose-and-limb structure, with a skeleton driving drawn characters, provided the conceptual model for skeletal animation's bone hierarchies. Computer systems formalized the animator's armature-and-pose practice as jointed skeletons.
Skeletal animation influenced by Manipulator kinematics and dynamics in Robotics
Manipulator kinematics and dynamics provided the mathematical apparatus of joint chains and transformations that skeletal animation used to compute limb positions. Robotics formulations of articulated chains were taken up by character animation systems.

Computer Animation is a planet in Nousfarer
Its city has a tower for each of these 7 frameworks. Land, walk in, and climb one concept at a time.
Land on Computer AnimationRelated fields
Fields its ideas are linked to
- Physical Simulation 5 links
- Animation Theory 3 links
- Computer Graphics 2 links
- Robot Kinematics Dynamics 2 links
- Robot Manipulation 2 links
- Sports Biomechanics 2 links
- Virtual Reality 1 link
- Computational Physics 1 link
- Programming Paradigms 1 link
- Rehabilitation Biomechanics 1 link
- Robot Motion Planning 1 link
- Robotics 1 link
More in Computer Graphics
- Computational Imaging
- Computer Graphics
- Game Technology
- Geometric Modeling
- Graphics Rendering
- Physical Simulation
- Virtual Reality
See where Computer Animation sits on the Map of Human Knowledge