XmasVibe ✕ VIS PLASTICA

Drawn with Vector Brushstrokes, Not Diffusion Pixels

Today, most people associate "AI art" with diffusion models (like Midjourney or DALL-E) that hallucinate smooth, synthetic pixel grids in seconds. These 197 holiday works come from an earlier, quieter moment in computational art history.

Between 1 December 2021 and 6 January 2022, a tiny headless mini-PC hummed continuously in the corner of Dylan Banarse's (2ne1.com ↗) studio running mathematical vector synthesis. Instead of generating pixels, the system starts with a blank canvas and places hundreds of mathematical Bézier vector curves - iteratively adjusting stroke paths, thicknesses, and translucent pigments until a neural vision model recognizes the holiday subject.

The Gesture of a Machine Learning to Sketch

Because Vis Plastica is restricted to a finite number of continuous vector strokes, it cannot rely on photographic trickery. Like a child drawing with oil pastels or a Fauvist painter working from memory, it must capture the essential gesture of Five Golden Rings, Seven Swans a-Swimming, or a glowing Christmas tree through bold, overlapping marks.

Each day in December 2021, the computer completed between 4 and 12 independent drawing sessions - producing 138 Advent studies and 59 Twelve Days studies. Because the optimization starts from random stroke trajectories every time, each study is a genuinely one-of-a-kind mathematical artifact, rendered directly from its vector curves at 4× resolution (896 × 896 px).

100% of Profits to War Child UK via Teemill

Dylan Banarse takes zero commercial profit from this project. Every garment and print is manufactured on demand by Teemill on the Isle of Wight using solar power and 100% GOTS certified organic cotton. Teemill acts as Merchant of Record (handling payment processing, VAT, and plastic-free Royal Mail fulfillment), and 100% of net store royalties (£5.58 to £7.49 per item) are donated to War Child UK (Registered Charity No. 1072216) under a formal Commercial Participation Agreement via Work for Good.