A short, looping generative animation: 15,000 colored points on a black background continuously reorganize themselves into five recognizable painterly compositions that smoothly morph into one another. Visually the style nods to pointillism — the image is made entirely of individual colored points — but this is a simplified, decorative version of the technique: real pointillism by Seurat and Signac relied on an optical color-mixing theory (complementary colors placed side by side so the eye blends them at a distance). Nothing like that is calculated here — each point’s color is taken directly with a small random jitter, and points are simply layered with transparency. So “dot-based generative graphics” is a more accurate description than “reproducing the technique of specific painters.” What’s depicted (5 scenes): 1. Poppy field — a green meadow scattered with scarlet poppies with dark centers and thin stems; a pale sky strip above. 2. Sunflowers — several large sunflowers with discrete yellow-orange petals, a brown center disc, curved green stems and leaves. 3. Lavender field — rows of lavender spikes (dark green base → lilac tip) receding toward the horizon, clouds and sky above. 4. Water lily pond — dark water with a moonlit reflection, a regular grid of round lily pads with dark edges and light centers, bright star-shaped flowers with yellow centers among them. “Water lily pond” is a well-known, widely reused motif (most associated with Monet), but that’s just a thematic reference to the subject, not an attempt to reproduce his brushwork — Monet was an impressionist, not a pointillist, and his technique is different. 5. Tulip field with a windmill — colorful rows of tulips (red, yellow, orange, pink, white, purple) receding toward the horizon, the silhouette of a Dutch windmill with blades, sky with clouds. How it’s made (technical — verified against the code): • Tech stack: pure HTML5 Canvas 2D and JavaScript, no external libraries or CDNs — a single self-contained file. • 15,000 points are stored in typed arrays (Float32Array) holding each point’s position, RGB color, and radius. • Each of the five scenes has its own generator function that computes a target position and color for every point: • “lobed” shapes for flower petals and canopies — points distributed in a circle with the radius modulated by a sine function; • discrete petals placed around a circle at fixed angular steps (sunflowers); • vertical “spikes” that taper toward the tip with a base-to-tip color gradient (lavender, tulips); • a regular grid with local jitter (lily pads), two separate point clusters for the windmill (a tapering trapezoid body and blades). • Scene transitions: each point’s position and color are linearly interpolated (with ease-in-out) from its current state to the target over 1.6 seconds; each painting is held for 5 seconds. • Micro-animation: each point has one of three oscillation modes — a faint background flicker, a “wind sway” for stems/petals (a traveling wave along X), and a ripple for the pond’s water surface. Amplitudes are in the 0.012–0.05 range of world units — genuinely subtle motion. • The black background isn’t a historical detail, just a contrast choice — colored points read brighter and shapes stay legible against it. • The placement of flowers, rows, and clouds in each scene is regenerated on every loop (using Math.random() with no fixed seed), so the image differs slightly from run to run while staying recognizable.