numeric
photography problem set 9 this week i set out to improve my problem-6 applet, PUTTYPIX, which had been designed to suggest or simulate silly putty. puttypix worked by assembling a small set of atomic images according to the rules of a rudimentary shape grammar. although the results achieved a moderate degree of seamless continuity, they nonetheless suffered from an overly-rectilinear suprapattern imposed by their own grid. in this week's amendments, i decided to dispense with the shape grammar altogether and turned instead to a promising if computationally-expensive model based on implicit surfaces, or metaballs. my first
attempt at a revision using metaballs, PUTTYPIX2,
was, despite substantial changes to both its interaction and graphics
models, frustratingly similar in percept to the original puttypix. in
order to achieve a more malleable facsimile of blobbiness, i sensed
that i would have to additionally toss out my method of straightforwardly
abutting constituent images. instead, i sought a wholly pixel-based
way of combining metaballs with photographic source material. I am indebted
to my colleagues rich devaul and ben fry for their brainstorming assistance
in the course of developing my second attempt, BLOBBY. |