Table Of Contents
- The Early Days of Animation
- The Importance of Personality
- “The Mind Is The Pilot”
- Art Classes for Animators
- The 12 Principles of Animation
- Lasseter’s Updates
- In Summary
- “Readable” versus Optimized?
- Discovering Personalities
- “Shadowy” Creatures
- Animation Exercises With Vector
Carlos Baena helped to create the character designs of Cozmo and Vector. Mayfield Robotics hired Doug Dooley to guide the designs of Kuri. Alonso Martinez has been building his own “Mira” robots. Baena, Dooley and Martinez have all been involved in significant character design and animation for Pixar. It’s also difficult not to see some influence of Pixar’s “Luxo” and “Luxo Jr” on the design of Jibo.
There’s no denying Pixar’s international success in animated movies. So what might Pixar character designers and artists know that applies to social robots?
In 1987 “Principles of Traditional Animation Applied To 3D Computer Animation” was published by John Lasseter (who later became Pixar’s Chief Creative Officer). Lasseter updated Disney’s “basic principles of traditional 2D hand drawn animation” for 3D computer based animation to “demonstrate the importance of these principles“. Those “fundamental principles” as he called them were first published in “Disney Animation: The Illusion of Life”. Widely considered a bible for those serious about animation “The Illusion Of Life” felt like the right place to start to understand more about animation and what it might have in common with social robots.
The Early Days of Animation
In the early days of animation during the 1920s there was little money to be made and as a result there was barely any appetite for innovation. Some animators even considered it a “chore” to have to move characters around the screen. Many preferred to get straight to setting up the often “preposterous” gags and hope to get an audience reaction. If a gag found even moderate success with a few laughs from the audience it was regularly reused.
Walt Disney, however, saw enormous potential in the new medium of animation. Walt believed that “successful animations” were “real things that had feelings and emotions and thoughts” so that audiences would care about what happened to the characters rather than see them as the butt of every joke. He felt there was “no life and very little warmth” until an event could be seen through a character’s eyes.
The Importance of Personality
Walt noticed that Disney Studio’s animations weren’t really taken seriously nor did they make any significant impression on audiences until they started to create personalities. He felt that “for a character to be real, he must have a personality” and that it was part of the scaffolding that helped guide the rest of a character’s design.
It was heavily stressed that they should not be one dimensional imbecilic characters or there purely for comedic purposes. Part of the believability of a personality was that people could identify with the character and what they’re going through on screen. His animators were often reminded that there was plenty of “entertainment” to be found in personalities that were derived from “qualities common to all individuals“. The teams assigned to an animation were required to put enormous efforts into the tiny details and refinements needed to make the personality “an individual – unique and entertaining“. Interestingly, predictability was their ultimate test for a “well defined” personality i.e. “can we imagine how this character would believably react in any situation?”.
As they continued to strive for believable personalities, the animators realised how much motion was required. In order for the characters to “stay alive” on screen they couldn’t be completely stationary nor was it enough to say a character’s personality trait existed, “it must be shown“. They discovered that personalities were “defined more by their movements than their appearance“, that they are “revealed not so much in speeches as in mannerisms” and how easy it was for a character’s believability to be lost if a single motion or action was not right for their personality.
“The mind is the pilot”
“In most cases, the driving force behind the action is in the mood, the personality, the attitude of the character – or all three. Therefore, the mind is the pilot. We think of things before the body does them” – Walt Disney
While trying to figure out how to show a character’s internal state the animators stumbled upon an idea from an acting class: thought processes and emotional states can be communicated non-verbally through changes of expression. This turned out to be quite a breakthrough for the animated medium. A change of expression could be shown in a variety of ways such as a “single held drawing, a simple move, through gestures, body movements or full actions“.
It was also crucial that the change of expression was seen to come from the mind of the character. The “illusion of life” broke down if the audience felt they were seeing the result of a gimmick or even an artist showing off a trick. Fred Moore noted that audiences “loved to see the drawings move and the characters think…we should always let them see the characters think!“.
In their earlier animations the artistic demands were much simpler, often limited to gestures an audience might interpret as anger or annoyance and “a kind of fright that wished it could be ‘fear‘”. Walt wanted to tell deeper stories with characters that displayed much more complex emotions like “love, dejection, hate, jealousy, concern and fear“.
Art Classes for Animators
Walt had set a very high bar in his demands for engaging, believable and life-like personalities that would provide the types of complex characters and stories he wanted to create. He was also very aware of the limitations in contemporary animation and felt most of it “lacked weight“. There were few attempts to recreate anatomy or realistic movement, especially human movement. For example, a character’s feet were simply drawn moving up and down if they were walking, if they ran their feet moved up and down faster. If they were reaching for something, the length of their arms would grow or shrink depending on how far away the object was they were reaching for.
There were exceptional artists and animators on his staff but Walt was unhappy with the general inability to reliably create high quality animations. He invested in art classes (that ran at night, after work of course!) for all of his animators in the hopes of developing a systematic approach to teaching animation that would bring all his current and any future animators up to the desired standard. Additionally, he encouraged his animators to support one another by sharing their knowledge. If an animator showed a particular ability for something Walt wanted everyone to benefit from their expertise. This was quite uncommon at the time. Job scarcity forced many in the animation industry to keep their processes a secret.
Those classes, taught by Don Graham at the Chouinard Art Institute, often consisted of watching, re-watching, discussing and analyzing frame-by-frame footage of things like people walking, sitting, and lifting heavy objects. The animators also studied the motion of live models and animals. Walt wanted them to understand real life motion so well that they could use that knowledge to make imaginary characters believable.
Leo Salkin summed up their ultimate goal after observing a riveting moment in an acting class “What made that work? Why did that happen at that moment? That golden moment is our goal. That is what we must understand and recreate in our own medium“.
Each discovery made in the art classes built up their “visual literacy” and helped to identify what was or was not life-like in their own animations. The processes involved in creating “visual symbols of communication” started to become more refined, more reproducible and they were taught to every new animator who joined the studio “as if they were the rules of the trade“. Those rules would become the “12 Principles of Animation“.
The 12 Principles of Animation
While there are 12 separate principles they are used in combination to create believable life-like motion in animated characters. They’re applicable to the overall movement of a character just as much as to their constituent parts like their facial expressions, clothing or bodies. I was interested to look at them in a little more depth and to see how they might also apply to social robots.
1. Squash and Stretch
The Disney animators considered this their most important discovery. Most animated character movement at the time was stiff and very rigid. Gestures were somewhat recognizable but had a limited sense of “vitality” and were often more like “a diagram of the action” than life-like animation. In reality almost all living things will show considerable changes in their shape as they move. Muscles contract and relax as facial expressions change. A person’s legs swell and stretch as they jump. The extreme positions from the squashed to the stretched pose and vice versa help to give that living feeling of fleshiness and flexibility.
For a social robot that uses facial expressions this would be an important principle to pay attention to, for example, eyes that squint slightly with a smile or a furrowed brow that widens in shock or surprise. And while not obviously applicable to metals, plastics and glass used in robot bodies, a hinged or articulated body could certainly show the contrasting changes of shape with its ability to squash down and stretch up (like the Pixar Lamp).
Almost all organic movement has an element of anticipation. It is “the natural way for creatures to move, and without it there would be little power in any action“. For example a dog crouches before it leaps, a foot draws back before it kicks a ball, eyes will stare at a cup as a hand raises up before reaching out to pick it up. Early animations were full of abrupt unexpected movements. This wasteful style meant audiences often missed punchlines or were left confused because they were unprepared for any of it. The Disney animators divided actions into three parts:
- The preparation for the action
- The action itself
- The termination of the action
Anticipation isn’t for communicating why a character is doing something. It is there to prepare the audience for what is about to happen so that they don’t miss the action. When anticipation is done correctly the audience will know what to expect before it actually happens.
This is a very important principle to consider for social robots. Their purpose is to interact with people. If a robot immediately began an action it’s likely a person would be caught off guard or miss the action entirely leaving them confused and possibly disinclined to want to interact with the robot any further.
Staging is considered “the most general” of the principles because it covers so many areas – “an action is staged so that it is understood, a personality so it is recognizable, an expression so it can be seen, a mood so that it will affect the audience”. When staging action the key is to only do one thing at a time. It should be staged “in the strongest and simplest way” so the audience always knows where they should be looking and what they should be looking at.
For a social robot to be understandable then this principle should be at the core of every action it performs. Without it, interactions would be unstructured, drawing a person’s attention to the wrong things leading to confusion and making the robot seem unpredictable or faulty.
4. Straight Ahead Action and Pose to Pose
This is the only principle that actually offers advice on drawing. Almost all hand drawn animation is made up of Straight Ahead Action or Pose to Pose.
Straight Ahead allows for a lot of creativity and is worth considering when a “fresh, slightly zany look” is needed. The animator is aware of the main plot points of the scene to include and they start with the first drawing continuing straight ahead until they’ve drawn enough to reach the end of the scene. Some proponents of this method felt “the animator should be as surprised as anyone at the way it comes out“. Straight Ahead suits “wild, scrambling actions” but there may be issues with the animation becoming unfocused, lacking variation in timing and inconsistencies in the shape and volume of the character through the scene.
Pose to Pose offers more control and structure. The motion is planned in advance before any drawing happens so that each pose relates to the others maintaining size, position and relationship to the staging. Pose to Pose helps to ensure the character’s personality and actions are “handled carefully for the maximum clarity, appeal and communication“.
It is possible to combine both approaches so that the Straight Ahead action doesn’t get too out of control. A rough path of the character’s likely progress through a scene is sketched out before any real drawing is started. The sketches aren’t part of the finished animation acting only as a guide to reduce some of the possible issues already mentioned with Straight Ahead Action.
Straight Ahead Action is what we would refer to as Procedural Animation these days. This would require the robot to be entirely responsible for generating its own animations in real time and technically with little pre-planning. It seems like an extremely advanced and complicated task to achieve while also ensuring the animation is readable. It might work well for very short animations that provide a little novelty, like idle animations of characters in a computer game, rather than large more complex animations that need to communicate intention.
Pose to Pose would seem like a better choice for a social robot. The important poses are already planned out by an animator to ensure they’re readable and the right information is communicated. It’s also possible to iterate on these animations much more easily than Straight Ahead. Pose to Pose runs the risk of becoming boring as the same animations are repeated over time but “texture” can be provided by adjusting timings, intensity and some of the other principles.
5. Follow Through and Overlapping Action
This principle is about how to handle the 3rd part of an action – “the termination of the action“.
Follow Through is concerned with making a character seem like they exist in a World with physics and bio-mechanics. In the same way there is the anticipation of a foot drawing back to kick a ball, the foot will follow through the action after making contact with the ball.
Overlapping Action (or simply “Overlap”) supports the follow through after an action. Bodies do not stop all at once. Instead their constituent parts often stretch, twist and contort. Things like appendages, hair, clothing and flesh move at different speeds and how they slow down will be proportional to their weight.
Aside from the physics of it, a person doesn’t reset themselves after they complete an action before starting into the next one. It would be incredibly unnatural to do so. Actions overlap each other and maintain a continual flow often highlighting an underlying thought process.
I think the really important parts of this principle for a social robot is the importance of correctly handling and communicating the termination of actions. It doesn’t necessarily mean they have to kick footballs or throw punches realistically but their facial expressions shouldn’t abruptly appear and disappear, their limbs and bodies shouldn’t jolt in and out of poses and critically to maintain the illusion of life-like motion, the termination of the action should be proportional to the anticipation that preceded it.
6. Slow In and Slow Out
When living things move they start and come to rest in a progressive way. There’s acceleration at the start of a movement as a person slows out of their current pose and a gradual deceleration as they slow in to the next pose or come to a stop. Earlier animations lacked this appearance and sense of natural movement because all drawings were equally spaced in an animation. The solution was to use more drawings closer together to slow an action down.
Funnily enough this principle is about how to avoid creating abrupt “start stop” mechanical or robotic movements. Smooth motion is very natural looking and pleasing to the eye which would help to make a robot’s movements more life-like, make their actions easier to understand and encourage future interactions.
Through their art classes the animators discovered “the movements of most living creatures will follow a slightly circular path“. Those arcing paths are not only pleasing to the observer but are “the most economical route a shape can follow when moving“. An immediate impact of this was a change to how characters walked. Prior to this characters “popped up and down like mechanical gadgets” whereas now the desired life-like approach meant the feet should arc over at the top of the step and arc under at the bottom position.
It would make sense for a social robot to incorporate this style of movement. Firstly, linear motion is considered very mechanical and unnatural and secondly, as arcs are a very common movement pattern in the natural world it is one that people are already familiar with making a robot’s motion easier to understand.
8. Secondary Action
A Secondary Action is something that is caused by the main action of a scene. It helps to make the main action more believable without distracting from it. It is something else the character is doing in addition to their main action. A nervous character might constantly look around while drumming their fingers on the table. A sad character will have a sad facial expression but also their posture will be slumped and they’ll move slightly slower than usual.
Secondary Actions are behavioural patterns or tics rather than the physical motion that Follow Through or Overlapping Action correspond to. A social robot could leverage extra secondary actions to help it strengthen what it is communicating in as clear and understandable a way as possible.
Timing is considered one of the most important principles. Proper timing “gives meaning to movement“. It refers to how quickly something in a scene moves or how long it stays still. In those very early cartoons timing simply meant moving characters fast or slow. The Disney animators learned that subtly varying the timing of movements was the difference in a character being perceived as “lethargic, excited, nervous, relaxed“.
This is just as important for animated characters as it is for robots. Improperly timed interactions would be confusing. Timing is also one variable that provides “texture” as mentioned in Straight Ahead and Pose to Pose. Texture would give some variety to interactions with a social robot over a long period of time if there are animations that would be regularly repeated.
Exaggeration helps to develop “the essence of an action“, to accentuate personalities and moods so that they are more convincing and easier for the viewer to understand. If a character is meant to be sad make them sadder. If they’re meant to be angry then make them furious. It is a principle on its own but in reality it’s applied to many of the others like Squash and Stretch, Secondary Action, Timing and Appeal.
For the animators it provides an opportunity to inject a lot of fun and dynamism by creating contrasts in poses like the squash of a curious facial expression that then stretches incredibly wide in shock. Of all the principles this one seemed to cause the most confusion amongst the Disney animators because Walt would demand realism only to criticise them because “it was not exaggerated enough“.
To me exaggeration is about making sure what an action, a mood or a personality are meant to communicate is blindingly obvious. Exaggerated facial expressions, poses and movements would help to make a robot’s actions and internal state more understandable and convincing. It could also help to inject some fun and novelty in interactions if required.
11. Solid Drawing
While Solid Drawing is about making things that are 2D images appear to be 3D, the Disney animators also learned the importance of avoiding too much symmetry or “twins” in a character’s appearance. Humans rarely exhibit perfect symmetry. People often shift their weight more to one leg than the other when they stand, facial expressions are typically asymmetrical and we rarely position our hands symmetrically either. Some slight asymmetry helps to maintain the character’s life-like appearance.
Avoiding symmetry might be something to be aware of for a limbed robot, a robot that is representational of a head and/or a torso that their head is rolled slightly and animated facial expressions may also be more life-like if they incorporated slight asymmetry.
Appeal doesn’t mean “cuddly bunnies and soft kittens“. It may be last in the list because it is an overarching principle as not only is a design or appearance appealing, movement can also be appealing and is something to always keep in mind when working with the previous eleven principles. According to the Disney animators “Appeal” is anything a person likes to see – “a quality of charm, pleasing design, simplicity, communication or magnetism” – and will hold a person’s gaze while they appreciate what they are seeing. Appeal is very subjective and often difficult to be precise about. In fact Disney and his animators were able to be much more specific about what was not appealing: “weak drawing“, “a drawing that is complicated or hard to read“, “poor design“, “clumsy shapes“, “awkward moves” and “attempting too much refinement “.
Clearly this is a very important one for a social robot. It is one of the most difficult to address too. A character in an animation only has to be appealing for the length of time they are on screen. A social robot has to get beyond the “Novelty Effect” and continue to be appealing so that people will want to interact with it regularly over long periods of time.
In his paper, John Lasseter highlighted that the principles “mean the same regardless of the medium of animation” but some like Pose to Pose require a different approach in 3D. As it was redundant he removed the “Solid Drawing” principle and considered “twinning” a part of “Appeal” adding that “if each part of the body varies in some way from its corresponding part, the character will look more natural and appealing“.
He did provide a wonderful quote from Walt Disney discussing “Overlapping Action” that reinforces the importance of characters thinking and how thoughts are what tie movements together:
“For example, the mind thinks ‘I’ll close the door – lock it – then I’m going to undress and go to bed’. Well, you walk over to the door – before the walk is finished you’re reaching for the door – before the door is closed you reach for the key – before the door is locked you’re turning away – while you’re walking away you undo your tie – and before you reach the bureau you have your tie off. In other words, before you know it you’re undressed – and you’ve done it in one thought, ‘I’m going to bed’“.
The principles are based upon the hard earned knowledge that “Disney’s Nine Old Men” and other Disney animators gained in those early years of studying and producing believable life-like animations capable of non-verbally communicating the internal thoughts and emotional states of Walt’s complex characters. They helped Disney Studios create some of the most memorable characters in some of the most highly regarded animated motion pictures of all time.
They’re also a testament to the astuteness of Walt Disney who saw the potential in animation and invested in developing processes that would create the “successful animations” he wanted. As animator Les Clark commented “Today it may seem simple to us; at the time it wasn’t.“
Social robots are designed to communicate and interact with people. They may do this through their own computational models of biological processes or canned pre-programmed actions and responses. Regardless of the source, their actions need to be believable and readable to the people they’re interacting with. Humans use and interpret body language – posture, gestures, facial expressions – all the time. So who better to help create a robot’s “body language” and ensure it reinforces what the robot is trying to communicate than the likes of Pixar animators and character designers? They’ve proven time and time again they know how to use the “visual symbols of communication” detailed in the 12 Principles to create the “illusion of life” through animation and appealing personalities.
The whole notion of life-like animation and robotics raises some interesting thoughts.
“Readable” versus Optimized?
As the Disney animators noted, “Very few living organisms are capable of moves that have a mechanical in and out or up and down precision“.
If we want them to seem life-like then social robots need to move in a way that is “readable” and not necessarily in the precise or efficient (in terms of considerations like power consumption) way we might expect from motors and electronics.
For example applying the principle of “Anticipation”, a robot turning its head should move its eyes first, or, the principles of “Arcs” a grasping robot’s arm should not move linearly but follow an arc and as with humans the wrist leads the fingers when reaching out for something. This begins to get extremely complicated.
Remembering back to Walt’s example of someone preparing for bed and how one action blends into another through an overall thought process. A robot trying to move in a life-like way also has to do incredibly difficult things like recognize objects around it, motion plan in an unstructured environment like a home, generate speech utterances etc. and do so in fractions of second.
The original Disney test that a well defined personality should be predictable hasn’t really changed. Nowadays, Pixar suggests imagining how a character would react if they were “trapped in an elevator“. And as with the 12 Principles a lot of what the Disney animators discovered about personalities might seem obvious to us now. The below “Flypaper sequence” is from 1934’s “Playful Pluto” (animated by Norm Ferguson) and is considered a milestone in personality animation because it was “the first time a character seemed to be thinking on the screen“.
How does a social robot display its personality ? Or should that be how might a person discover a robot’s personality? Characters in animations have devices like the plot, antagonists and their own character arc to display the attributes of their personality. I think the “Three Laws Of Robotics” would have very limited application or appeal in real life!
I really liked the proactive approach Baena and his team used by making Cosmo a “pathological show-off” and let it communicate its personality. It echoes what the Disney animators discovered that personalities are “revealed not so much in speeches as in mannerisms“. The robot moving around by itself forces people to use their inherent tendencies to anthropomorphize it which further reinforces the “illusion of life“. Users can engage with it further through interactions like asking it questions or playing games.
Taking inspiration from similar devices that Disney Studios used, I would imagine that interactions can be treated like scenes that are created around daily routines, regular events and other special events as story points. These would give the robot the opportunity, like Cosmo, to be engaging, to ask questions and drive the “plot”. The difficulty of a predictable personality remaining engaging over a long time (much longer than an animated movie for example) is still an issue. A mixture of “texture” in regular animations as well as occasionally novel interactions for special events could help to counter that.
Do we really need to create social robots that look and move just like humans? It seems sensible to copy our existing “social interface” – our bodies and faces – it is very expressive and we know how to interpret it so there is nothing new to learn. It may strengthen a social robot’s ability to communicate clearly by reducing the potential for misinterpretation.
The “Uncanny Valley” hypothesis from the 1970s is well known in relation to artificial characters and humanoid robots. I was intrigued to read about a similar problem the Disney animators ran into back in 1937 during the making of “Snow White and the Seven Dwarfs“. Always looking to improve their animation skills the animators started to have actors act out scenes for them and the camera film was transferred frame by frame to photostats. Whenever the animators matched the actions of the actors in frame by frame detail the result was incredibly unsuccessful. They found “the moves appeared real enough, but the figure lost the illusion of life” and “it was impossible to become emotionally involved with this eerie, shadowy creature who was never a real inhabitant of our fantasy world“.
One of the main reasons for the failure of that approach was that real people move with too much complexity to draw and too much subtlety to support an animated personality believably. Woolie Reitherman summed it up as “the art of animation lends itself least to real people and most to caricatures and illusions of a person“.
Seven years later, 1944 Heider and Simmel conducted an experiment where participants were asked questions about a very basic animation. Most people perceived these incredibly simple moving geometric shapes as animate beings with intentional and goal driven behaviour.
If people can attribute such complex behaviour to simple shapes that lack gestures and facial expressions then it’s likely they’ll do the same with something that looks like a human and naturally expect it to behave in complex human ways. That sort of robot or artificial intelligence does not currently exist and when expectations aren’t met then only the differences and what’s lacking will be seen. It would be akin to asking a celebrity impersonator who they can’t do impressions of.
That isn’t to say roboticists should completely abandon highly anthropomorphic designs. The best way to test an understanding of something is to attempt to model it. However, I do feel the higher the level of correspondence to a human the higher the level of expectation and the lower the bar for failure. Given how much humans naturally attribute to movement focusing more on the goal of life like movement might be a better goal.
Animation Exercises With Vector
I would absolutely recommend reading “Disney Animation: The Illusion of Life” if you can find it. As someone who comes from a technical background it was really useful to understand the problems with early animation techniques, how to describe motion, how characters were developed and how the studio figured out to bring all the parts of a feature length animation (story, sound, dialogue etc.) together. At times I was amazed at the issues they had to solve but that was because I had taken it for granted that animation was always the way I’ve seen it on television or in the cinema.
There are many tests and exercises that beginning animators are given. I tried to pick a few that were possible with the Vector robot I own. These were all created programmatically within the constraints of the current version of the Vector SDK. I learned an invaluable amount from the “Illusion Of Life” and wanted to try applying it. I also learned that an animation may never actually be 100% complete, there’s always another little adjustment or slight variation but like blog posts they’ll never be finished if you keep tweaking them!
A typical animation exercise is a character leaping across a large gap or an obstacle. Instead I animated Vector backing up and charging forward at the highest speed the motors would allow. There is a lot of anticipation for the action in the slow back up like a spring compressing (some slight implied squash), the head lowers as a sprinter does in their blocks and as Vector accelerates forward the head lifts a little (some stretch to go with the squash) to try to exaggerate the burst of force.
This was the most difficult of the animations but definitely the most enjoyable one to attempt. There’s plenty of anticipation and timing to get right in the early stages of the pre-sneeze (if that’s a word!) as Vector slowly backs up slightly farther each time while the head goes back with the arm and eyes narrow to a close. The motors aren’t strong enough to send Vector racing backwards so the force of the sneeze is heavily exaggerated in the reactionary wide-eyed head clearing shake at the end.
Turning With Anticipation
The intention was to use a quick blink and the eyes lead the movement helping to anticipate the turn. Unfortunately, SDK doesn’t give complete control over the eyes and creating blinks programmatically was difficult.
Building on the previous turn and trying to add in some more anticipation by almost taking a small step backward and picking up the arms before turning.
When a person turns they dip their chins and the movement of their head follows a slight arc. These animation exercises are typically done with a torso or head and shoulders so I tried to imagine Vector’s entire body was a person’s head. The timing was difficult to perfect on the slight arcing motion and if the ability was there to control it more, Vector’s actual head would have straightened up as the turn ended rather than slightly after it.
And finally, for a laugh, I did an exaggerated “robotic” turn with a straight ahead stare. Without the anticipation (the eye movement) preparing you it is impossible to predict which way Vector will turn.
“Meet the Smartest, Cutest AI-Powered Robot You’ve Ever Seen“, https://www.wired.com/2016/06/anki-cozmo-ai-robot-toy/
“Interview: Doug Dooley, Kuri Animation Designer“, https://www.heykuri.com/blog/interview-doug-dooley-kuri-animation-designer/
“Alonso Martinez’s 3D-Printed Animated Robots!“, https://youtu.be/0vfuOW1tsX0
“Luxo Jr. – Trailer“, https://youtu.be/lkKf9DWmR04
Lasseter, J. (1987). “Principles of traditional animation applied to 3D computer animation.” Computer Graphics, 21(4), 35–44.
Leo Salkin – https://www.imdb.com/name/nm0758494/
Heider, F., & Simmel, M. (1944) An experimental study in apparent behavior. The American Journal of Psychology, 57, 243-259