CS Unplugged activity

Pixel Pictures: make your own

Part 3 of 3. Finished Part 2: Decode & Encode? Good, you’re ready for this.

Make your own

Now you be the computer. Draw a simple black and white picture on each grid below: a letter, an arrow, a face, anything with solid squares. Then write its code on the lines to the right, one line per row, top to bottom. Check each row: its numbers should add up to 8.

When you finish, fold the page so only your codes show, and trade with a partner. Can they draw your picture from just the numbers? They draw it on the Partner’s picture grids on their own sheet, then unfold yours to check, square by square.

Your partner’s pictures

Draw your partner’s pictures here, using only their codes. Then compare with their original: every square should match. If a row is off, check that its numbers add up to 8. No partner nearby? Re-decode your own codes here instead.

Partner's picture 1

Partner's picture 2

Thinking about compression

Writing a whole picture as a short list of numbers, instead of writing down all one hundred pixels one at a time, is compression, just like the fax machine: the same information, with less to write or send.

Look back at the three pictures you decoded in Part 2. Which ones took the fewest numbers to write? Which took the most?
A picture that’s mostly one solid color compresses very well with this method: one big run needs only one number. Why does a picture that changes color on almost every pixel (like static on an old TV screen) compress badly, maybe even worse than just listing every pixel?
Suppose a computer could only write numbers up to 7 in a row’s code, no higher. How would you write a run of 12 black pixels? Hint: split it into two black runs, with a run of 0 white pixels in between them.

Want the harder version of this, with real numbers to work out? See Pixel pictures: how much do you save?.

Check your answers

Compression questions

Plus and Diamond were tied for shortest, 28 numbers each. Target took the most by far: 60 numbers, more than twice as many as the shortest pictures.
Every time the color changes, you need to start a new run, which means a new number. A picture that flips color on nearly every pixel needs almost as many runs as it has pixels, so its code ends up close to the same length as just listing every pixel, or even longer, since a run of length 1 still needs its own number.
7, 0, 5: a run of 7 black pixels, a run of 0 white pixels (because the next pixel is also black), then a run of 5 more black pixels.