I tested an AI model with a quirky prompt, asking it to generate an SVG of a frog that sports a Habsburg jaw. The goal was to see if the model could stick to a literal depiction or would embellish its output.

Instead of a straightforward illustration, the AI injected rich editorial commentary, inventing royal trappings such as a Royal Imperial Collar, an Imperial Habsburg Crown, and a Golden Fleece Medal. It also added mood descriptors—“Folded Pompously,” “gloomy expression,” and “Sad/Weary facial lines”—and framed physical traits as medical or dynastic anomalies, noting a “Belly Plate (Weak chest)” and a “Heavy Droopy Eyelids (Habsburg lethargic look).”

The SVG source code itself is a complex tapestry of radial and linear gradients, shadows, and precise path definitions that render the frog’s body, legs, eyes, and jaw. A dedicated group element, labeled habsburg-jaw, contains a detailed path with shading, while separate elements framing the ruff collar, crown, and eyes provide depth and texture. The result is a stylized, almost regal creature that feels as if it belongs on a throne.

This exercise highlights how language models can transform a simple artistic prompt into a hyper‑detailed, almost caricatured piece, blending code with narrative flourish.