Please make a copy of this text file, fill in your answers, and place
them in your Project1 handin directory. Please call the file
"questions.txt". Please do not delete the questions from the file.

This assignment is due Thursday, October 19th, 9:30 am. However, we
will accept late handins of these questions up until 5pm on October
20th without penalty.

Please provide SHORT answers to the following questions. We will tell
you how long your answer should be. If you feel the need to write more
than this, you are probably not saying things as concicely as you
should. (and when we give a line length, we mean "standard" width
70-80 character lines).

You are not allowed to discuss this assignment with anyone
else. (consider it a takehome quiz).

You are to do this assignment AFTER you have handed in your Project.


Name:


Login:

1. Quantization and Brightness - the BW command

There are 3 "common" ways to convert to black and white: quantize each
pixel (just round to black or white), ordered dithering, and Error
Diffusion. (call these Q, O and E).

You could convert to grayscale first (using perceptual weighting of
RG and B) and then Quantize, or you could simply have your quanization
pick Black or White. (call the former G and the latter P)
In an email, I mentioned that one of those options is correct.

This means there are 6 possible ways to have implemented the BW
command QG, OG, EG, QP,OP, and EP.

1.A. Which did you implement (2 letters):


1.B Describe an image that would allow us to tell which of the six
  options an assignment implemented (4-5 lines):



1.C Describe the result of applying QG to this image (e.g. what would
  you look for to tell that the assignment had implemented QG) (2-3
  lines):


1.D For OG:


1.E For EG:


1.F For QP:


1.G For OP:


1.H For EP:




2. Provide a script that creates a checkerboard of opaque white
   squares and transparent (black) squares using only the COPY, SOLID
   and COMPOSITE commands. You should make the overall image size
   30x30, the "checkers" 10x10, have the upper left corner be white,
   and use as few SOLID commands as possible.


   (this script should be between 6-12 lines long)



3. You have implemented at least 1 matte generation algorithm. Pick
   the one that you have found works the best and answer the following
   questions:

   3.A What is the operation? Describe it in 2-4 lines


   3.B Describe a non-trivial problem that it would work perfectly 
       (or at least extremely well on) (2-4 lines):	      


   3.C Describe an image that it would give noticable (bad) artifacts
       on, no matter how much you tweaked the parameters. 
       By "bad artifacts" we mean places where someone using the image
       in a compositing operation will see that pixels will not have
       the correct alpha values.
       Describe these artifacts for this bad image (2-4 lines):



4.A. Describe your method for doubling (2-3 sentences)


4.B. If your algorithm can add high frequencies to an image, describe
     an image on which it would add high-frequencies to, and explain
     why (including how this addition of high-frequencies would be
     visible). 
     Otherwise, say "it never does." (either 1 line of 4-5)


4.C. If your doubling algorithm can remove high frequencies from an
  image, describe an image that it would have this behavior, and
  explain why (including how this loss of high-frequencies would be
  visible). 


4.D. Suppose that we assume that our image was created by correctly
  sampling a band-limited "real-image." 

  It is possible to create a size doubling algorithm that given an
  image that met the above sampling criterion produced an image twice
  the size that neither added nor removed high frequencies.
  
  Explain why (e.g. give an outline of the method that would be used). 

  HINT: it requires a two stage process.
  HINT 2: consider what it means for the input to the algorithm to be
          "sampled correctly."

  (WARNING: This is a difficult question. Even with the hints.)


5. Suppose we had a version of the assignment that implemented the
   half operation by simply taking every other pixel (e.g. pixel 0,0
   2,0, 4,0 0,2, 2,2, 4,2, ...), but had the other basic operators
   defined correctly (e.g. crop, composite, solid)


   Describe how to implement a half operation with good sampling using
   the available commands (e.g. the bad half command, and the other
   operators).
   (you may either choose to give a short description, a script, or
   both)