About

A word cloud maps term frequency to visual weight. Each word's rendered font size is proportional to its occurrence count f within the source text, normalized against the maximum frequency f_max. Misreading frequency distributions leads to skewed emphasis in presentations, reports, and content analysis. Common stopwords ("the", "and", "is") are filtered using a 150-entry English stopword list to prevent noise from dominating the output. This tool uses an Archimedean spiral placement algorithm with axis-aligned bounding box collision detection to pack words without overlap. It approximates layout density but does not guarantee mathematically optimal packing. Pro tip: paste at least 200 words for a visually dense cloud. Single-paragraph inputs produce sparse results.

Formulas

Font size for each word is computed by linear interpolation between a minimum and maximum pixel size, scaled by its relative frequency.

size_i = size_min + f_i − f_minf_max − f_min ⋅ (size_max − size_min)

The Archimedean spiral used for placement positions each word candidate at coordinates:

x = a ⋅ θ ⋅ cos(θ), y = a ⋅ θ ⋅ sin(θ)

Where a controls spiral tightness (default 1) and θ increments by 0.1 radians per step. Collision detection uses AABB overlap: two rectangles collide when x₁ < x₂ + w₂ ∧ x₁ + w₁ > x₂ and the same for y axes.

Where f_i = occurrence count of word i, f_min and f_max = minimum and maximum frequencies across all words, size_min = 12px, size_max = 80px (configurable).

Reference Data

Stopword	Typical Frequency Rank	Category
the	1	Article
be	2	Verb
to	3	Preposition
of	4	Preposition
and	5	Conjunction
a	6	Article
in	7	Preposition
that	8	Conjunction
have	9	Verb
I	10	Pronoun
it	11	Pronoun
for	12	Preposition
not	13	Adverb
on	14	Preposition
with	15	Preposition
he	16	Pronoun
as	17	Conjunction
you	18	Pronoun
do	19	Verb
at	20	Preposition
this	21	Determiner
but	22	Conjunction
his	23	Pronoun
by	24	Preposition
from	25	Preposition
they	26	Pronoun
we	27	Pronoun
her	28	Pronoun
she	29	Pronoun
or	30	Conjunction

Frequently Asked Questions

The generator removes approximately 150 common English words (articles, prepositions, auxiliary verbs) before counting frequencies. This prevents high-frequency grammatical words from dominating the visual layout. You can disable this filter in settings if analyzing raw token distributions.

Words shorter than 2 characters are excluded by default. Additionally, if many unique words exist, the generator caps at the top 150 by frequency to maintain visual clarity. Words with frequency of 1 in a large corpus may fall below this cutoff.

Words are placed using an Archimedean spiral starting from the canvas center. The highest-frequency word is placed first (center), then each subsequent word spirals outward until a collision-free position is found. This produces a roughly circular, dense layout.

Enabling rotation allows words to be placed at 0° or 90° angles. Rotated words often fit into gaps that horizontal-only words cannot, increasing packing density by roughly 20-30%. However, vertical text reduces scan speed for readers. Use rotation for aesthetic posters, disable it for analytical presentations.

The tokenizer splits on whitespace and punctuation, which works for most Latin, Cyrillic, and CJK-separated scripts. However, the stopword list is English-only. For other languages, disable the stopword filter to prevent English words from being incorrectly retained while target-language stopwords pass through.

The export matches the canvas dimensions, which default to 800×600 pixels. On high-DPI displays, the canvas may render at 2× internally but exports at the logical pixel size. For print use at 300 DPI, this covers roughly a 6.8×5.1 cm area. For larger prints, increase the canvas size before generating.