React component for shadcn/ui
ASCII Decode Text
A React text animation for shadcn/ui where a headline first appears as ASCII art of its own letter shapes, then a decode front sweeps across and condenses the art into real, crisp, selectable type, with a pointer lens that re-encodes the text under it.
Example
Features
- The noise is the headline itself. The real text is drawn once at its
measured line boxes, and each small cell of a monospace grid takes the
character for how much ink falls in it, from
.:-=+*#%@. You can read the word as ASCII art before it decodes. - A decode front sweeps from the start, the centre, or random points, easing in and out. Cells flicker through neighbouring characters just before they resolve, and the real text is revealed behind a narrow soft-edged mask, so letters arrive crisp rather than as a blurry cross-fade.
- The art is tuned per theme: on dark backgrounds the glyphs are drawn lighter and with a little less ink, so they don't glare.
- After decoding, a decoder lens follows the mouse, pen, or a sideways touch drag and re-encodes the type under it. It opens in 140 ms and the cells behind it heal back to type over 300 ms. On the text, the lens replaces the mouse cursor.
- The art overlays the real text exactly, with the same line breaks, width, and height, so nothing shifts when it finishes. Multi-line headlines, any font, letter-spacing, and RTL text work, because the art is sampled from the browser's own layout.
- At rest only the real text is painted: the canvas is cleared and no frame loop runs.
- An optional, very soft type tick as the front crosses each letter, and a light haptic when a touch lens starts.
Installation
pnpm dlx shadcn@latest add @tentui/ascii-decode-text
Usage
import { AsciiDecodeText } from "@/components/ascii-decode-text";<AsciiDecodeText>Decode the signal</AsciiDecodeText>children must be a plain string: the art is rasterised from it. The
headline renders as an h2 in the inherited font, text-4xl font-semibold
by default, with display leading and tracking per size and balanced wrapping. Change the element with as, the scale with
size (sm, md, lg, xl), and anything else with className. The ASCII
cells use your font-mono in the current text colour.
Triggers
By default the headline shows as ASCII art and decodes once, when 40% of it
is in view. trigger="mount" decodes right away, and trigger="hover"
decodes on hover, focus, or tap, re-encoding first if it is already decoded.
<AsciiDecodeText trigger="mount" direction="center">
Hello, world
</AsciiDecodeText>Controlled and imperative
With trigger="manual", decoded controls the state: true decodes and
false sweeps it back into art. A ref exposes replay(), encode(), and
decode() in every mode, and onDecoded fires each time a decode finishes.
const headline = useRef<AsciiDecodeTextHandle>(null);
<AsciiDecodeText ref={headline} as="h1" size="xl">
Every interface starts as noise
</AsciiDecodeText>
<button onClick={() => headline.current?.replay()}>Replay</button>Look
density sets the cells per cap height (4 to 12, 7 by default) and ramp the
characters from empty to full ink. tone="muted" draws the art at about 60%
of the text colour, and dust sprinkles faint characters around the letters
while encoded.
<AsciiDecodeText density={9} ramp=" .oO0@" tone="muted" dust>
Low signal
</AsciiDecodeText>Interaction
| Input | Action |
|---|---|
| Hover with a mouse or pen | Re-encodes the text under a soft lens that heals behind it |
| Drag sideways on a touch screen | The same lens; vertical swipes still scroll the page |
| Press and drag with a mouse | Selects the text; the lens steps aside |
Hover, focus, or tap (trigger="hover") | Re-encodes, then decodes |
While the lens sits on the text, it is the cursor: the mouse cursor is hidden there and comes back over the empty parts of the box, while you press to select, and while the text is decoding. If the pointer is already resting on the text when a decode finishes, the lens opens under it.
The lens appears only once the text has decoded. Its radius is 0.6em,
clamped to 24-64 px; set it with lensRadius, or turn it off with
lens={false}. It is always off with trigger="hover" and under reduced
motion. Keyboard focus does nothing special unless trigger="hover", because
the lens is decoration.
Sound and haptics
With sound, a very soft tick plays as the front crosses each letter, at
most every 35 ms and 24 times per decode. It never plays before the page has
had a user gesture, for the lens, or in a hidden tab. Sound is off by default.
A single light haptic plays when a touch lens starts, on devices that
support it. Turn it off with haptics={false}.
Accessibility
The real text is always in the DOM, from the first paint. While it shows as
art, it is visually masked rather than removed, so screen readers read one
heading with the real text and never any ASCII gibberish. The canvas is
aria-hidden. The text stays selectable and copyable, even while encoded.
With reduced motion, or animated={false}, there is no sweep and no flicker:
the art fades out and the text fades in over 240 ms once triggered, one after
the other so they never overlap at half strength, and the lens is off (so the
cursor always stays visible). The art's lightest cells are drawn at 72% of the
text colour (65% on dark themes), which keeps 3:1 contrast for large text on
the default themes.
In right-to-left text, direction="ltr" still means "from the start of the
line", so the front starts at the right. The art is sampled from the real line
boxes, so bidi text keeps its shaping and order.
API reference
FAQ
How do I install ASCII Decode Text with shadcn?
Run npx shadcn@latest add @tentui/ascii-decode-text after adding the
@tentui registry to components.json (installation guide).
The CLI copies ascii-decode-text.tsx into your project and installs
web-haptics and @rexa-developer/tiks.
How is this different from a text scramble or decrypt effect?
A scramble swaps each letter for one random glyph of the same size. Here the noise is an ASCII rendering of the actual headline: many small cells per letter, each picking its character from how much ink of the real glyph falls in it, so the word is readable as art before a front condenses it into type.
Can screen readers read the headline while it is animating?
Yes. The real text is always in the DOM and only visually masked, so assistive
technology reads one heading with the real text from the first paint. The
ASCII canvas is aria-hidden, and the text stays selectable.
Does it hurt performance?
The text is rasterised once per layout, and again on resize or when fonts load. A single frame loop runs only while decoding or while the lens is moving, the canvas is capped at 2x pixel density, and at rest nothing is painted or animated.
Does it respect reduced motion?
Yes. With prefers-reduced-motion or animated={false}, the art simply
fades into the text over 240 ms, with no sweep, no flicker, and no lens.
Is ASCII Decode Text free to use?
Yes. ASCII Decode Text is free and source-available: the shadcn CLI copies its source into your project, and you can use it in unlimited personal and client projects under the TentUI License.