To gain a deeper insight into just who this audience of fathers might be, I needed to carry out some research. From all the in depth information this webpage provides, I isolated the fact that the majority of children are born to men between the ages of 20 and 44.
If we take 5 years old to be a good age when children should begin developing their reading skills, then their fathers could be aged anywhere between 25 to 50 years old, which is a large spectrum. I also nee to bear in mind that British fathers come from a broad range of cultural and ethnic backgrounds, and that while the brief cites low income families as a particular concern the problem of poor literacy is not limited to this group. My storyboard will need to have universal appeal so that it can have a lasting impact upon all of these groups in the hopes of achieving lasting change.
I initially had quite a bit of trouble getting the audio to play properly at 12 fps, the specified frame rate for this task. It turned out that the solution was to change the working unit of time (Window-Settings/Preferences-Preferences-Settings) to 12 fps as opposed to film 24 fps, rather than just changing the playback speed in the Animation Preferences menu. Once the audio was successfully imported, I could begin to study and examine the sound clip, in order to break it down for the dope sheet. A dope sheet is an extremely useful reference tool for animators, allowing you to note down the frame at which certain sounds and actions should occur. The visual representation of the audio shown in the timeslider (see screenshot below) gives me an idea of when something is being said. However, it shows the background noise (of which there is quite a lot) and music as well as the main speaker's dialogue, so it is not always clear when a word begins or ends. Luckily, if you move the time slider a frame at a time it will play the relevant audio for that frame, allowing me to break down the dialogue into its individual sounds, so I know what phoneme to animate at each frame.
Another problem to overcome in this task is that a significant communicator in phonemes is the position of the tongue, which Moom is unfortunately lacking. Exposure of the teeth can also be a helpful indicator as to which sound a character is making, but initially I chose to remove Moom's large buck teeth, mainly because they are extremely ugly, but also because I thought they might obscure the rest of the mouth shape. However, without the teeth I have to concede a number of the phoneme shapes blended together, so I was forced to put the teeth back in. Consequently, so that the phoneme isn't hidden by the teeth, exaggeration will be all the more important; I will need to take recognisable mouth shapes and emphasise these in order to produce a clear lip sync. However, the 12 fps timeline means that there simply aren't enough frames to include every phoneme, so some will simply have to be left out. Luckily, the audio recording I am working with ('but they'll never take...our FREEDOM' from Braveheart) has quite a few soft syllables which makes it easier to identify the most suitable phoneme. For example, Mel Gibson doesn't really pronounce the 't' at the end of 'but' or the 'k' at the end of 'take'. It was quite hard however to convince myself that these syllables could be left out, that they didn't need to be included.
There are also quite a few different controllers that can be animated on Moom's face to produce a range of expressions, and the mouth alone has 6 (7 if you include the jaw) controllers. Unfortunately this meant there were a few occasions throughout this task when I forgot to set a keyframe for a controller that didn't need adjusting at that particular frame (autokey only works if you change a controller's value). Consequently the automatic in-betweens had facial elements moving when I needed them to remain stationary. However, it was a good learning experience and I can be prepared for this issue when producing my 'Acting Up' animation for this module.
Despite how negative my experience of this study task may sound, and how many frustrating little issues I encountered, I actually very much enjoyed this task. I found it very rewarding to see this not very appealing character develop a personality and begin to come to life.
Study Task 3 required us to choose 5 emotions from a given list, photographing ourselves in poses which communicate these emotions, and finally using these photos as reference images for positioning Moom.
I had already begun this study task before my interim crit presentation, but after the feedback I received on incorporating more exaggeration into Moom's poses I decided to revisit the emotions I had already completed and rework them. Rather than just look at Moom from the angle shown in the photograph, I found it useful to rotate the character as I worked, to avoid having to make major alterations later on. This helped to ensure that the pose would look good from any angle.
As I have already mentioned, I am not a natural performer, and was consequently a little uncomfortable posing for the reference images, meaning my poses were not as strong as perhaps they could have been. So, rather than stick rigidly to the poses I had assumed for the photographs, there were occasions where I felt the emotion could be better conveyed by modifying the pose, even if it would not be physically possible for a real person to hold that pose. A prime example of this is in the 'anticipation' pose, where Moom leans forward at an angle it would be impossible for him to sustain without falling over. However, bending the rules of gravity, physics and anatomy is what helps make animation such an effective communicator of mood, tone and emotion.
The Uncanny Valley is a term which refers to a dip in a person's level of comfort when observing a figure with almost, but not quite, natural looking human features. Although originally applied to robotics in the 1970s by Doctor Masahiro Mori, the phrase is increasingly used in 3D computer animation.
The most uncanny animation is often found where motion capture (or the more detailed Performance Capture, which records subtler human movements and expressions) has been involved, where hyper-realism has been striven for at the expense of creating characters that the audience can empathise with. The artists are literally driven to distraction by what ever advancing technology can achieve. Some oft-cited examples are Robert Zemeckis' films The Polar Express and Beowulf.
The animated counterpart simply doesn't capture the Hanks' expression and vivacity; he looks dull in comparison.
The hypothesis is that as the realism of a character increase, so too does audience empathy. Up until a point; beyond this point, empathy drops steeply and can even lead to feelings of revulsion and hostility. Yet it is uncertain at what point on the scale of realism the negative reaction begins. For some, even characters in Pixar favourite Toy Story provoke this reaction.
Somewhat paradoxically, the detractors of The Polar Express claim the characters are 'doll like', 'awkward' and even 'robotic', although thanks to performance capture they are as human as possible beneath their CGI shells. However, this does support one theory as to why the Uncanny Valley exists. The more immediately apparent it is that a character is nonhuman, the more noticeable their human qualities become, while the more apparently human a character is, the more noticeable their nonhuman, or alien qualities become. Hence the popularity of anthropomorphism, an also perhaps why another motion capture creation, Gollum, escapes much of the uncanny criticism. His emaciated and considerably devolved physical states renders him adequately 'nonhuman'.
A particularly unnerving child from The Polar Express.
Released in the same year as The Polar Express, Brad Bird's The Incredibles uses similar 3D CGI animation, but without the motion capure, to create more stylized and cartoonish characters. The Parr family correspond more with how we are used to animated characters behaving thanks to years and years of 2D animation, with almost Tex Avery levels of exaggeration in both their physical proportions and movements.
Even in non comedic and cartoonish genres, animation works best when capturing the essence of natural movement rather than copying it verbatim. Observing recognisable quirks and traits and incorporating those into a character's performance is a more effective method of imbuing them with personality than slaving over their appearance.
Another by product of feature animations which use performance capture are the bizarre looking characters which manage at the same time to be both highly familiar and extremely alien. Beowulf is a prime example, starring an unnervingly youthful and athletic Ray Winstone (similar to the digital Jeff Bridges in Tron Legacy), dead-behind-the-eyes Angelina Jolie, and a whole host of other famous faces in shiny CG form.The design team even went so far as to use technical trickery to incorporate some of Winstone's facial features onto the dragon his character fathers. Yet I for one can't help feeling that recreating the whole cast as virtual puppets did little to make audiences care or invest in the story. Would there not have been more vivacity in the characters if it had been filmed in live action? Or if the actors' performances and voices had been used as the basis for creating more appealing and stylised characters which were outside the realms of the Uncanny Valley?
After learning some basic modelling skills, the next step in developing our knowledge of Maya was to learn some basic animation skills. For Study Task 2, we were set a number of short tasks that would allow us to practice using different animation tools. The graph editor is perhaps the most important of these tools, as it allows you to adjust the automatic in between movement Maya applies to key frames.
The first, and perhaps most simple task, was to produce a turntable animation, showing a complete 360 degree rotation of an object (I choose to use Moom rather than either of my models from Study Task 1). All that was required to create this animation was the addition of a turntable camera to a scene containing the object you wish to rotate. So, with Moom selected (and in the Animation menu set) I simply selected the Turntable... option from the Animate menu. The number of frames and the direction or the turn can be adjusted in the Turntable... options menu.
The processes for the Bouncing Balls, Pendulum and Anticipation tasks were all fairly similar; after setting the key frames at the high and low points of each object's path of movement, the majority of the work was in the Graph Editor (found in Window - Animation Editors - Graph Editor). In the graph, the Y axis represents the attribute value and the X axis the time line/frame number. The graph can display a number of different curves, one for each keyed attribute of an object. To minimize confusion, you can select the relevant attributes in the window to the left of the Graph Editor and only those curves will be shown on the graph.
Before editing in the Graph Editor, in the Curves menu, select Weighted Tangents, then in the Tangents menu Free and Break the tangents, so that each tangent handle can be moved independently. If the tangents aren't broken, when one handle is moved the other will automatically move so that they remain symmetrical.
The main aim of using the Graph Editor with these task was to achieve a more natural looking animation by having the balls bounce quickly off the ground before slowing as they reach their peak in the air (the pendulum would similarly need to speed up as it enters a swing before slowing as it begins to climb again). There is no magic formula to this, just trial and error of adjusting the handles until you're satisfied with the result.
I based the red ball on a tennis ball (hard with a high bounce), while the green ball was based on a beach/volley style ball, although with more exaggerated squash & stretch.
The most intimidating of the tasks was undoubtedly the segmented pendulum, but thankfully the Graph Editor made this a manageable task. After setting the key frames for a normal pendulum swing (we were provided with scenes for each task with the correct controllers in place), in the Graph Editor the curve for each ball needed moving by a couple of frames. By doing this, the impression could be created that the swing was led by the top ball and carried through the rest of the pendulum.
Finally, to ensure that all of the elements keep moving in this pattern (and don't start and stop at different times), in the Graph Editor select Infinity from the View menu, and then in the Curves menu set both the Pre ad Post Infinity options to cycle, which should repeat the curves endlessly along the timeline.
Oddly, although the overlapping action task looked tricky I found the anticipation and bouncing ball ones more frustrating, as there was no fixed set of instructions to follow, just your own judgement and grasp of the principles of animation. The process behind the segmented pendulum was a little more formulaic, which made a nice change.
