In this assignment you'll experiment with some DCT code to get a sense of how the encoding works. First, you will need source and data files; go here to do that.

Once you have that running, do some exploration. I'll leave the details of this up to you; you should turn in a little report describing what you looked at and what you found (your grade will be based on how much you do and the quality of your analysis). Some questions you might consider are:

• At what quality level do you begin to notice significant degradation? Does that level differ among the sample images provided? If so, how do you explain it?
• The image files are simply sequences of byte values (read the comments for details), so you can create your own image files just using putc(). (You may need to change some constant values in the dct code if you do this.) Generate some simple image files (e.g. all the same color, simple stripes, other easy-to-generate patterns; study how well they compress compared to "real" images like the ones provided.
• For more advanced fun, study the format of compressed dct files (note that it's slightly different from the format we talked about in class) and "manually" create your own simple files -- you can use the bitio module to help.
• In tandem with the above, you could hack the dct code to output the dct matrix before it gets compressed, so you can get a sense of what the dct algoirthm does with various kinds of patterns.

(by the way, this is intended to be a little bit fun.)