SCI199Y: Computational Reality, Illusion and Deception
Assignment 2
Due: 11:59pm, Monday 8 November 2010
In our discussions of computer graphics and animation, I have often referred
to the concepts of models that provide a representation for something,
and simulation, that plays out one or more models in some way.
Think about this as we explore in this assignment a
particular kind of "realism" in
visual depiction. Our starting point will be a provocative idea that
tries to tie the "realism" of the depiction of human-like things
to the "emotional response" people have of observing such depictions.
I would expect that there will be a wide array of reactions to such
an idea. In fact, there is considerable
speculation on the topic already out there.
Your task will be to do some research, and to reflect on and to explain
your reaction.
Please read the assignment carefully, and we will discuss it in class or tutorial.
The "Uncanny Valley" is an idea or conjecture
first proposed by Dr. Masahiro Mori in 1970. The
Wikipedia entry
gives an overview of this conjecture, including citations to publications on
this topic, as well as an English translation of Dr. Mori's first
publication on the topic.
At the time, Dr. Mori was a researcher in robotics, or the study of
electro-mechanical devices that can perform operations
similar to cognitive and biomechanical activities
performed by humans or other animals.
In this assignment, with our background in computer graphics and animation,
we will explore the relevance of the
"Uncanny Valley" to visual depiction in animation, images and videogames.
The main aim, in fact, will be to get closer to what we might mean when we
use the word "realism".
-
Briefly define the "Uncanny Valley" as described by Dr. Mori and
the Wikipedia entry.
Provide examples of the evidence given for this conjecture by him and others,
making sure you cite your sources for such justifications; provide some
of your own examples that could be seen as supporting the conjecture.
Similarly, give an overview of the criticisms of this conjecture.
-
The "Uncanny Valley" was originally about the "realism" of
robots that somehow physically behave or look like us.
Recently, people have thought about the connection of the "Uncanny Valley"
to visual depictions of us, particularly in the context of images,
animations and videogames.
Let's see if we can make any sense of this.
To this end, read the discussion of this topic in the
Wikipedia entry,
including the article
"When fantasy is just too close for comfort" (http://www.theage.com.au, 10 June 2007).
I find the image in that article particularly interesting: is that image
really too close for comfort?
Some experimental work is emerging that is attempting to find
a connection between "anthropomorphism" and emotional responses such as
comfort, eerieness or fear.
For example, read the first few pages of the recent paper,
"Anthropomorphism Influences Perception of Computer-Animated Characters' Action"
by Chaminade et al. (see their
web page
for examples of the animations used in their experiments).
Now, a graph of "Uncanny Valley" looks like this:
By the way, an earlier version of the Wikipedia article used the following
much simpler graph that looked like this:
Look carefully at the similarities and differences of these two versions.
Now, write the best definitions you can of "human likeness",
"familiarity", and "realism" in the context of the first graph with
respect to visual depiction.
Note that I am not asking you simply to write out definitions you get from a
dictionary or Wikipedia, though you can use them as a starting point.
After writing up your own definitions, discuss specific aspects that you felt
had to be included in the definitions and those that you felt could be omitted.
For example, you might believe that absolute sub-microscopic physical accuracy
is not required for an image to be "realistic".
Discuss any shortcomings inherent to your definitions (don't be shy!).
In both graphs, there is an assumption that all these concepts
can be somehow measured or described as a single value.
Does this make sense to you in light of our discussions in class and
the definitions you have proposed?
As you think through this, look at some images I collected in a
sample portraits page .
Would you in broad terms assess those images differently according to your
definitions of "human likeness" versus "familiarity" (or "emotional response")?
Think about the animations you have seen,
or videogames you have played or that you know about.
Summarise your thoughts regarding the definitions you have proposed
and their applicability to the images, animations, and videogames you have
seen.
-
As mentioned earlier, the "Uncanny Valley" is a conjecture.
Please look up the word "conjecture" before going on.
In light of the above discussion, does this conjecture make sense to you for
any of the kinds of visual depictions we have considered so far (images,
animation, videogames)?
Do you believe the conjecture to be true or to have a grain of truth?
Is there an alternative, enhanced or more restricted form of the conjecture
that you would instead propose?
Is it simply a metaphor?
Is it completely wrong?
What do you make of the Chaminade paper in light of your views?
Marking Scheme
This assignment (like most of those in this course) is quite open-ended, and
there really is no wrong answer.
The quality, thoroughness and novelty of your justification and
evidence are key criteria.
Furthermore, the work done in each question accumulates and drives the next.
As such, the marking scheme will be slightly flexible, with Question 2
nominally counting for 50% of your grade and Questions 1 and 3 each counting
for about 25% of your grade.
The submission process for your assignment is the
the same as the last one.