Word Doodle – A generative engine that turns text into dense doodle art

Name: Word Doodle – A generative engine that turns text into dense doodle art
Availability: InStock
Author: koalux

by koalux·Mar 3, 2026·2 points·1 comment

Visit Project View on HN

AI Analysis

●●SolidEye CandyCozyRabbit Hole

Word cloud reimagined as animated doodle packing—relaxing to watch, collision-detected layout.

Strengths

•Lightweight vanilla JS with clever 70-attempt collision detection creates genuinely tight, organic packing.
•Watching words fill the canvas feels tactile and intentional, not mechanical like standard word clouds.
•Solid technical execution: frequency mapping + scaling + rotation creates visual hierarchy that reflects content.

Weaknesses

•Use case is purely aesthetic/decorative; no export quality or practical application beyond novelty.
•Word clouds already exist (Wordle, Tagxedo, raw D3 examples); doodle aesthetic is the only differentiator.

Post Description

I have always loved the look of dense, hand-drawn doodle sketches. The art that filled every part of a notebook page or paper with words and character drawings. Using word-doodle, I have recreated that feeling, including simulation of the process of filling the page but for text only and also emojis.

How it works: The engine uses a three-step pipeline to transform raw text into art: 1. Frequency Mapping: It analyzes word frequency to determine visual weight of each word. 2. Scaling & Rotation: High-frequency words get bolded and enlarged, with a 15% chance of vertical rotation. 3. Collision Detection: It uses a randomized coordinate generator with a retry loop (up to 70 attempts per word) to pack the canvas as tightly as possible without overlapping.

The Goal: Beyond just a word cloud, it is something relaxing to watch. It’s built with vanilla JS to keep it lightweight as possible.

GitHub Repo: https://github.com/j-ncel/word-doodle