oneshotlm
Compare all models on this prompt →

Flow-field particles · qwen3-coder

Using the pre-provided p5.min.js (global p5, instance or global mode), create a full-screen flow-field particle animation: thousands of particles advected by Perlin-noise vectors, leaving fading trails, with colors shifting over time.

⛨ sandboxed Open in new tab ↗
transcript · 4 messages
user

Using the pre-provided p5.min.js (global p5, instance or global mode), create a full-screen flow-field particle animation: thousands of particles advected by Perlin-noise vectors, leaving fading trails, with colors shifting over time.

assistant

I'll create a full-screen flow-field particle animation using p5.js with thousands of particles, Perlin noise advection, fading trails, and shifting colors. Let me build this as a self-contained index.html file.

write index.html (3521 bytes)
toolResult

Successfully wrote 3521 bytes to index.html

assistant

I've created a full-screen flow-field particle animation with the following features: 1. Thousands of particles (2000) moving according to Perlin noise flow field 2. Particles leave fading trails through a semi-transparent overlay technique 3. Color palette shifts gradually over time using HSB color mode 4. Responsive design that adapts to window resizing 5. Efficient grid-based flow field for performance The animation works by: - Creating a grid-based flow field using Perlin noise - Generating particles with random properties - Updating particle positions based on the flow field vectors - Drawing particles with fading trails using a semi-transparent background - Gradually shifting the color palette over time The visualization creates an organic, fluid motion as particles follow the noise-generated flow field, with colors cycling through the spectrum. Particles wrap around screen edges for a continuous effect. To run this, simply open the index.html file in a browser. The animation will automatically adapt to your screen size.

TAKE IT TO CODE

A single self-contained HTML file — copy it and open in any browser, or reuse the prompt with your own pi-agent run.

RUN

modelqwen3-coder
vendorQwen
harnesspi-agent
generated2026-08-01

EVALUATION

4/5 working

Flow-field particle trails with colorful, evolving noise-driven streams; continuous animation, no visible interactivity beyond time evolution.

Vision grade · anthropic/claude-sonnet-5

MEASUREMENTS

Raw readings from the capture, not a verdict — the grade above is the only judgement. Percentages are how much of the screen changed; the measure under-reports sparse motion on a dark background.

idle motion7.2%
on click5.6%
on drag5.4%
on wheel5.9%
on enter + space5.4%
on arrow keys5.7%
on w/a/s/d5.8%
frame spread7.5 / 255
console errors0
js errors none

METRICS

runtime17.2s
tokens in1.2k
tokens out1.2k
cached3.3k
cost / run$0.0018