Drawing a picture from memory using
Human Level Artificial Intelligence
Note: To make this website free to the public
please click on an ad to support my sponsors or you
can make a tax-deductable donation using Paypal
(click on the donation icon on the left).
This video shows a
robot drawing a picture from memory. There are no
sound in the video because I wanted to show the
viewers what the robot is thinking while making a
drawing. The flashing text and freeze frames are the
internal thoughts of the robot and not instruction
text for the viewers. These internal thoughts
describe the details of how the robot produce
intelligence.
My robot doesn't use:
planning programs/heuristic searches (used by MIT and
Stanford University), Bayesian's probability theories
for decision making, Bayesian's equation for
induction and deduction, semantic networks for
natural language understanding, predicate calculus,
common sense systems, first-order logic, rule-based
systems, genetic programming, or MACHINE LEARNING. By
the way, you can't use current AI methods to build a
robot that can draw a comic book. For example, the
IBM Watson can't be used to draw comic books. Those
researchers can spend 50 years programming
instructions into the Watson, their code foundation
disallows their robot to draw a comic book (or do
complex human tasks).
The robot learns to
draw pictures through a bootstrapping process. If the
robot wants to be a comic book artist, he needs to
learn art skills gradually. There are actually 3
videos made on this series. 1. Drawing a picture by
copying. 2. Drawing a picture from memory. 3. Making
a comic book page. Each video shows the gradual
learning the robot must go through in order to become
a comic book artist.
I want you to recall
all the art skills that you learned from school, from
Kindergarten to college. First, teachers teach
students to draw by copying real life objects. The
teacher would put an object, like a doll, on a table
and asks the students to copy what they see. The
teacher would tell students to look at shapes and
edges and draw lines on paper to represent that
object. Lessons about what to look at, where to look
at, and what lines to draw on paper are facts the
teacher teaches the students. Practicing drawing
different objects is part of the learning process.
Over the years, the students' brain uses personal
likes/dislikes to develop their own style of drawing.
Once the student can draw pictures by copying them,
then he is ready for the next lesson.
The second lesson
taught by teachers is to draw a picture based on an
image from your mind. This lesson is very similar to
the first lesson. Instead of copying an object from
real life, the student is copying an image from
memory. The student extracts an image of something he
wants to draw. Next, he looks at this image in his
mind and copies it.
The previous 2
lessons basically builds on itself to become more
complex. The third lesson is to draw pictures based
on storytelling. Learning the knowledge to become a
comic book artist is a very complicated thing. First,
the student has to know how to draw all objects, this
includes: human anatomy, cities, landscapes, cars,
highways, tables, animals, super heroes, faces, body
parts, things, and places. The student has to have
the ability to draw "all" objects and from "any" 360
degree angle. The student needs to have talent in his
artwork, which means that his drawings have to appeal
to a large group of people.
Next, the student has
to have knowledge about movie directing. He has to
have a movie playing in his head and the student
decides which scenes to show on paper. The comic book
artist usually works with a writer. The writer
provides a one page script to the artist. The artist
will read the script and design the page panels. In
each panel, the artist has to decide what camera
angle to use, what objects will be in each panel,
where to put the captions and so forth. There are
books written on how to make comicbooks and the rules
are vast.
Drawing a comic book
includes knowing knowledge about perspective,
shading, composition, angle shots, realism, anatomy
drawing, landscape drawing, environment object
drawing, linear story telling, and so forth. As
stated before, most knowledge for drawing is based on
practice. Practicing more will mean better skills for
the robot.
In the video, the
robot is trying to select an image from memory to
draw. He later decided not to extract an existing
image from memory, but to fabricate a brand new image
that isn't stored in memory. The robot's brain serves
as an image processor and it can process images. For
example, the robot's brain can take 2 or more images
and combine them to form a brand new image. In this
case, the robot needed an image of venom's costume
and an image of a character pose. He planned to draw
the character pose on the paper and superimpose
venom's costume on the character pose. This scene
shows the robot has the ability to fabricate images
from its mind. If the robot doesn't have an image he
is looking for in his brain, he uses his mind to
fabricate a brand new image.