Project Proposal

The Uncanny Valley is something many people are aware of, but they can never put a name to it. Do you enjoy pointing out the CGI in movies that try to trick you into believing the CGI is real? Or do you feel uneasy when you see something trying to mimic reality but it just isn't quite right? This is the Uncanny Valley. Essentially the name came from a dip in a graph created by robotics professor Masahiro Mori, that stated when a robot (CG character) is created to look lifelike but just misses the mark the viewer feels repulsed to look at it. If something is sufficiently not lifelike then the lifelike qualities will stand out to the viewer. On the other hand if something is “almost lifelike” then it is the unlifelike characteristics that stand out leaving you with a feeling of repulsion or dislike.

As The Uncanny Valley will forever exist it would be useless trying to get rid of it or ignore it. The best way to deal with it is to equip yourself with knowledge. This is what the project is all about. Equipping yourself and others with the knowledge to understand The Uncanny Valley so eventually it can be avoided. But first you must understand The Valley before trying to avoid it. I believe there are two sides of The Valley, Abstract (Up, The Incredible) and Realistic (Tintin, Avatar). In order to understand how to stay out of The Valley you need to understand the limits of those two spectrums. Learn those limits and how they work and in theory you could stay out of The Uncanny Valley.

I believe that the most crucial parts to staying out of The Uncanny Valley is to have believable movements. I agree that you need good visuals and convincing environments but the movement is the make or break area, the part that pushes it over the edge. Therefore I prepose to explore everything animation to do with abstract and realistic. To do this I will try my best to rig and animate both an abstract and a realistic characters. This practice, a long with research, will both educate and inform me on what makes and breaks the two areas. From there I will design and produce my own concise book of guides/tips to help the common animator stay out of The Uncanny Valley.

Abstract Rigging Workflow

1) Set Up Joints

Now the joints have been implemented into the mesh I can start to put in controls to help make it easier to move them. Controls are attached or constrained to the joints so that when you fish you only need to click the shapes surrounding the character instead of the joints themselves.

2) Add Controls To Joints

I have now added most the controls to the character. These controls have only been put on the Left hand and Centre of the rig and will be mirrored at the end of the process. After creating all the controls we now need to go into the hypershade in Maya and Do some "visual coding" as I like to think of it. This will allocate different controls to do different things so that when they are moved the joints react in a unique way predicted and chosen by you.

3) Add IK to Rig

Inverse Kinematics is where you can set a target from one joint to another so there is an almost elastic constraint between the two. Meaning that you can move just one joint and the leg will bend like a knee should or a tail will wag like you want it to. As well as adding IK to the rig you can also add pole vectors that help you twist the joints in a natural way.

4) Program Switches and Joint Restrictions in Hypershade 

In Hypershade you drag different nodes resembling the objects or actions to other nodes and link then together. For e.g. This finger controller will rotate but it will rotate 3 different joints evenly at the same time. And this can only be done by programming it to do so in the hyper shade. You can also use a controller to turn the visibility of other controllers on and off shown in the pictures down the bottom. This hand controller switches the arm between IK and FK by just changing a number as well as changing the controllers used. All using hyper shade.


Abstract Conclusion


After the question was asked "Why a beaver?" we realized that conceptually we had drifted and were looking at things too technically. Would we then try to make a realistic beaver? So we decided to move on from the beaver and create a second abstract character. This character would be the same as the realistic character showing the true differences between the two. Although the character was not bound to the rig I could still gather a conclusion on the abstract character by positioning the rig. In terms of movement and animation an abstract character is in most regards simple. Technically though the abstract is a nightmare. The more freedom the character has the more programming you need to do so the rig can perform. If you are truly worried about falling into the valley then an abstract character is a good path to take.

Abstract Animation Process

Since I had some extra time I decided to quickly animate the new abstract character so I could show some movement from them. It is extremely basic and there were a million problems with the weighting of the mesh but at least he moves...sort of. Unfortunately lost all its hair on the export.

Rigged abstract character with controls

Rowan Head Rub from Aaron Long on Vimeo.

Rowan Wave from Aaron Long on Vimeo.

Realistic Facial Animation

The face, more often than none, is the part of the animation that will drag it into the valley so quick you can't even say uncanny. Therefore I decided facial animation was very appropriate for the realistic animation experimentation. I decided the seven human emotion would be a challenging thing to recreate well. Underneath is a video of the seven emotions with a slowmo breakdown of each emotion. Under that are the individual emotions.

The 7 Human Emotions from Aaron Long on Vimeo.
I found the realistic model was much more enjoyable to animate. I think it may be because realistic is so much more familiar to me, to everyone. Its every day life and all you are doing is trying to recreate it. I would say it is also more self satisfying. With an abstract character you need to look to others for there opinion on wether it is funny or good, but for the realistic you can tell if its close to reality or not because you have been looking at it your whole life. In that respect it was easier in the way that you didn't need to be that creative to copy nature. The hard part is that copying nature is no walk in the park and is very time consuming. From these animations I compiled my findings and put them into a booklet I created.

The Uncanny Valley Book

In this book I have compiled some guidelines/tips I have found helpful, so animators can navigate through the realm of realism without falling into the Uncanny Valley. These guides/tips are by no means %100 fact but helpful processes to assist an animator on his journey to create lifelike animations. The Book includes things to look out for that put you in the Uncanny Valley, As well as things to do to help keep you out.

The Uncanny Valley Book Text

Things that puts us in the uncanny valley?

Eyes

• Lack of eye jitter (small, darty movements of two or three frames when someone is looking at something).
• Cross eyed or wall-eyed eyes; these eyes generally appear to be unfocused. You always need to be looking somewhere. (wax dummies, toy dolls)
• Wavering focus of the eyes; eyes not looking at a consistent target. (Not looking at something when you expect them to be looking. e.g. Not looking at someone when being spoken to.)
• Glassed over eyes. No normal eye looks like glass.
• If the eyes always move with the head when the head rotates, it looks robotic.
• When characters don't blink at all, it looks odd. The average person blinks once every five seconds.

Face

• Changes in facial expression that seem too slow, or are late.
• No one has a symmetrical face so “perfect” faces look wrong. a lot of people have one side of their face smaller than the other.
• Only one side of the face is moving; a common feature in stroke victims.
• When what they are saying does not sync up with there mouth.

Body

• Facial movements that are uncoordinated with what the body is doing.
• When there body moves too slow or too fast making it look un-natural
• When it is obvious your positions are going through key frame to key frame.

Tips to keep you out of the valley

• Micro Expressions. Small, quick and subtle expressions give a lot of personality and emotion to a character.
• Nostril Flares. Believe it or not those bad boys move! But don't over do them otherwise they will seem un-natural.
• Be careful when moving facial shapes individually. One movement on the face will cause another movement from something else i.e everything is connected.
• Don't over-enunciate. No one over does it when they talk. It is subtle and second nature. (A baby or someone speaking a new language will be different. Cartoony characters can sometimes get away with this)
• Don't forget swallows, gulps and neck tightens. If its a male his adams apple should move when he talks/swallows.
• Most peoples heads don't stay still for long periods of time. Don't just leave the head static, try and break the regularity of even movements.

• Rotate head to follow eyes, not the other way round. Remember the head moves because of the eyes.
• Look for key places to widen and flare the eyes. Eyes can also twitch in the lids and brows.
• Funny enough the Jaw can move side to side and in and out as well as up and down.
• Don't forget to breathe. Chest rise and falls are very important.
• Smile with the eyes as much as you do the mouth. In a fake smile only the zymgomatic major muscle (cheek) which runs from the cheekbone to the corner of the lips, moves. So you don't need to worry about the eyes in those situations. But it is up to you to judge wether to be real or fake.
• Finally make sure the expression on the face fits with the voice and what they are saying. If they are saying “I love you” they most likely wouldn't look angry.

Contextual Statement

The Uncanny Valley is something many people are aware of, but they can never put a name to it. Do you enjoy pointing out the CGI in movies that try to trick you into believing the CGI is real? Or do you feel uneasy when you see something trying to mimic reality but it just isn't quite right? This is the Uncanny Valley. Essentially the name came from a dip in a graph created by robotics professor Masahiro Mori, that stated when a robot (CG character) is created to look lifelike but just misses the mark the viewer feels repulsed to look at it. If something is sufficiently not lifelike then the lifelike qualities will stand out to the viewer. On the other hand if something is “almost lifelike” then it is the unlifelike characteristics that stand out leaving you with a feeling of repulsion or dislike.

Some people say the Uncanny Valley cannot be beaten like Angela Tinwell, senior lecturer in games and creative technologies at Bolton University. "We've come up with the notion of the 'uncanny wall. It suggests a viewer's discernment for detecting imperfections will keep pace with technology. As the human likeness increases, so the viewer becomes ever more discerning of little flaws. So when a new, more "realistic" character like Tintin comes along, they won't be rated as more humanlike – previous humanlike characters will actually sink lower into the valley. It's cognitive dissonance; we refuse to be tricked by technology."

Although I agree with Tinwell that the viewer becomes more discerning of little flaws as technology advances, I don't believe that the viewer cant be deceived into thinking CG is reality. For example in Tron Legacy (2010) the young Jeff Bridges was a CG character and fooled many viewers. I believe this is because of three elements working seamlessly together. Aesthetics, Movement and Environment. The Visuals matched the rest of the film style so you didn't question his character. The Movement was motion captured so that every detail was relayed onto the CG character. And the environment matched the character so the audience wouldn't question the placing.So as it has obviously been achieved it isn't impossible...just difficult.

As an animator I wanted to focus more on the movement side of the combination because I believe that is where the personality and emotion comes from in a character. I believe if animators and film makers were more aware of the Uncanny Valley and what brought characters into it then it would be easier to avoid falling in or letting your audience drag you in. I explored both sides of the Uncanny Valley, Abstract (Up, The Incredibles) and Realistic (The Polar Express, Tin Tin, Avatar) to understand the differences in workflow and restrictions between the two. I believe that through my findings it will help inform and improve my quality of work and understanding of animation. Through case studies and research, along with practical experimentation, I have derived a variety of guides/tips to help inform the animator of what puts a character in the Uncanny Valley and what to do to stay out. These guides/tips are by no means %100 fact but helpful processes to assist an animator on his journey to create lifelike animations.

In the second semester I wish to use these tips, along with the findings my peers discovered in my group, to create something that will deceive/trick the audience into believing or not noticing a CG character thus proving 1, the guides/tips work and 2, It is possible to deceive a well educated technological audience.