Animals Made from 13 Circles (2016)

AI capabilities and the 13‑circle constraint

• Some see the piece as a natural benchmark for AI: a small, well‑defined search space (13 circles, ~39 real parameters) producing rich, recognizable forms.
• Discussion asks what would need to change for LLMs to “explore” such spaces directly, rather than only describing solutions in words.
• Others note GPT‑4 already solves related tasks (e.g., procedural TikZ/SVG drawings), and suggest search‑based methods (evolutionary algorithms, MCMC) guided by vision models like CLIP could handle “13 circles” constraints well.
• There’s a side debate about whether everything must be viewed through the AI lens, with some pushing back on the AI obsession and others defending it as a natural curiosity.

Related mathematical and algorithmic art

• Links to Schmidhuber’s 1990s “low‑complexity art” and to Fourier epicycles (drawing arbitrary shapes with rotating circles) show precedent for “everything is circles” constructions.
• Kempe’s universality theorem and parametric “elephant fitting” are cited as analogies: a few parameters or mechanical linkages can approximate complex shapes.

Process of constructing the animals

• Commenters speculate on workflow: sketch curves first, then retrofit circles; or start from circles as construction lines, erasing and refining.
• Linked tutorial explains counting one circle per curve while sketching, then adjusting by adding/removing/moving circles digitally.
• Several connect this to foundational drawing pedagogy: building forms from circles/spheres, cubes, cylinders; “construction” drawing; and training the eye to draw what is seen rather than symbolic shortcuts.

Constraints, aesthetics, and low‑complexity art

• Many praise the strong visual clarity and personality emerging from severe constraints (13 circles + boolean operations over regions).
• This is framed as “low‑complexity art,” similar in spirit to bytebeat music. Constraints are said to foster creativity and aesthetic coherence, with parallels drawn to logo design, architecture, poetry, and experimental drawing setups.

Discoverability, nostalgia, and side threads

• Some lament that this kind of independent creative web content feels rarer or harder to find amid modern “slop,” noting search results dominated by Pinterest/Reddit before reaching the original.
• Others share related circle‑based resources (Twitter logo geometry, geometric compass art, Japanese family crests) and imagine animations, CAPTCHAs, or parametric “Animal” classes built on 13 circles.