Old news, but there’s nothing self-referential about a formula that draws all possible bitmaps of height 17, where the bitmap is encoded by the gigantic number n (not displayed in the image!) that defines the y coordinate where we choose to start the graph. Tupper himself didn’t even call it self-referential.
I was just going to ask whether one could port a full-fledged quine to this "platform", and then your Quora answer went on to exhibit one (!). Awesome attention to detail on your part as well as the constructor's!
I showed this to my teacher once. He looked at the equation and cracked it in a second. He even said the guy who made the formula probably used simple logic tricks to create it.
See, if you take that giant number for n and convert it to binary, you will get a long stream of bits. Take every 17 of those and draw them as an individual column of pixels. 0 is white and 1 is black. That's it. You'll see the "pixelated" plotting of the equation (but maybe mirrored or upside-down. Not sure).
My teacher showed me how you can use floor and mod (heavily used in Tupper's equation) to shave off bits from both sides of the sequence of bits that represent an integer. I have to ask him again, sometime. That was pretty dope.
I had a file called tupper.py that would let you draw something and then it would convert it to the corresponding n for Tupper's Self-Referential Equation. It would simply loop in the proper direction through your monochromatic drawing and annotate 1 if a pixel was painted and 0 otherwise. The result was the binary representation of the n where you would find your drawing.
For more details on how it works, see my Quora answer: http://qr.ae/TbcJFK.