Loading
Preparing your generator page
We are loading content and tools so everything is ready to use.
We are loading content and tools so everything is ready to use.
Turn any text into glitch text you can copy and paste, in 10 styles from Zalgo to static and raw binary. Set the glitch level, reroll the pattern, and clean it back up when you are done.
For horror-themed styles, try the Creepy Text Generator or browse the Text tools.
Last updated: October 5, 2026 · Published: 2026-03-24 · Updated: 2026-10-05
22 visible characters
Full glitch
s͈ͣ̽░y̺̦ͤ-s̖̐͒y͚͎͒́-sͭ̆҉̢͇y̿̀҉̡s̭̏ͮt̺̄e̯̬͒ͤ͞m̡̎ͫ g̗̭̐̒͟l̖̈ͥ͡▖i̱ͣt̶̹ͭ̇c̥̗̉ͭh̤ͨ͂ ḏ̅e͚̒̀͜▞t̞͓͛ͭ͜e̶̱ͫç̶͇̉̇t͈͈͊eͯ͟ͅd̤̑͛
31 visible · 116 code points · 118 of 280 on X
All 10 glitch styles
Glitch marks never stack more than 30on one letter, the limit in Unicode's Stream-Safe Text Format. Screen readers may skip or read out marks and symbols, so keep anything important in plain text.
A glitch text generator makes ordinary text look like a digital error. It stacks combining marks on letters, drops static blocks between them, swaps letters for the replacement character �, makes words stutter, or shows the raw bytes behind the text. "glitch" becomes ģ͚͆̋l̳͖̊ͦ-g̺̓ͅ░ḽ̸͉́͐-g̢͊̏ḽ̶̍i̮ͭ̚͡t̼̜̾c̡ͭh̭̏ͮ or gli▞tc▚h.
Because the result is plain Unicode text, not an image or a font file, it pastes into usernames, bios, captions, and chats.
Every pattern comes from a numbered seed, so the same settings always give the same result, and Zalgo marks never stack more than 30 on one letter. Clean up reverses the effects. Everything runs in your browser.
Type your text
All 10 glitch styles update as you type. Names, gamer tags, and short captions work best.
Set the glitch
Drag the glitch level from 1 to 8 and click New glitch to reroll the pattern.
Copy and paste
Copy one style or all ten as a numbered list. Each style shows its code points and how much of X's 280 limit it uses.
Each example is "glitch" on the default level. The last column shows whether Clean up gets the original back.
| Style | Example | How it is built | Code points | Weight on X | Reversible |
|---|---|---|---|---|---|
| Plain | glitch | Ordinary ASCII letters. | 6 | 6 | — |
| 1. Full glitch | ģ͚͆̋l̳͖̊ͦ-g̺̓ͅ░ḽ̸͉́͐-g̢͊̏ḽ̶̍i̮ͭ̚͡t̼̜̾c̡ͭh̭̏ͮ | Stutter, static blocks, and Zalgo marks layered at half the glitch level. | 47 | 47 | Yes |
| 2. Zalgo glitch | g̺̭͔̓ͮͭ͢l̸͎̲ͣͥȉ̱͚̾ͧͧ͡ţ͍̪͆ͫͧ̚͞c̡̺̥͚ͭ̈́̔hͨ͛҉̡̥̼̺ | Combining marks stacked above, through, and below every letter. | 48 | 46 | Yes |
| 3. Scanlines | gl̴it̷c̴h | Strokes and slashes drawn through letters with overlay marks U+0334 to U+0338. | 9 | 9 | Yes |
| 4. Static | gli▞tc▚h | Shade and block characters like ░ ▒ ▓ █ dropped between letters. | 8 | 10 | Yes |
| 5. Corrupted | g�itch | Letters swapped for �, the Unicode replacement character shown for unreadable data. | 6 | 7 | Replaced letters are lost |
| 6. Stutter | g-g-glitch | Words that skip and repeat their first letters, like gl-gl-glitch. | 10 | 10 | Yes |
| 7. Case flicker | GlITch | Letters that flip between capital and lowercase at random. | 6 | 6 | Original capitals are lost |
| 8. Leet | 6l17ch | Classic 1337 digits: 4 for a, 3 for e, 1 for i, 0 for o, 5 for s, 7 for t. | 6 | 6 | No, 1 and 0 can stand for several letters |
| 9. Binary | 01100111 01101100 01101001 01110100 01100011 01101000 | Every UTF-8 byte of your text as 8 binary digits. | 53 | 53 | Yes |
| 10. Hex | 67 6C 69 74 63 68 | Every UTF-8 byte of your text as 2 hexadecimal digits, like a hex dump. | 17 | 17 | Yes |
Binary is the longest: 53 characters for six letters, because each byte takes eight digits and a space. Corrupted turns glitch into g�itch, and Clean up gives back gitch, because the replaced letter is gone.
The level controls every style at once. Here is Zalgo glitch "glitch" at three settings.
| Level | Example | Marks | Weight on X |
|---|---|---|---|
| 1 of 8 | g͙̓͡l̦ͤi̸͎ͥtͩ͢ͅc͚ͭh͈͆͟ | 16 | 22 |
| 4 of 8 | g̺̭͔̓ͮͭ͢l̸͎̲ͣͥȉ̱͚̾ͧͧ͡ţ͍̪͆ͫͧ̚͞c̡̺̥͚ͭ̈́̔hͨ͛҉̡̥̼̺ | 42 | 46 |
| 8 of 8 | g̨̺̻̰̓ͮͭͧ͆̉͜͡l͎̖̱͚͉͙͍ͭ̽ͩ̏͟͟ͅi̡̲̗͕̺ͩͥ̋ͬ͌̂ͭ̚t͊ͨ͛̚҉̡̡̥̼̺͚̞͔c̻̘͔̞ͭͤ̒̽͛ͣ͜͞h̟̜̹ͭ̄͊ͩ̾́͋ | 74 | 80 |
Static uses shade and block characters from Unicode's Block Elements range, U+2580 to U+259F. The light, medium, and dark shades ░ ▒ ▓ stand for 25, 50, and 75 percent fill, and the original IBM PC already had them at bytes B0, B1, and B2 of code page 437.
| Character | Code point | Used in |
|---|---|---|
| ░ | U+2591 | Static, Full glitch |
| ▒ | U+2592 | Static, Full glitch |
| ▓ | U+2593 | Static, Full glitch |
| █ | U+2588 | Static, Full glitch |
| ▚ | U+259A | Static, Full glitch |
| ▞ | U+259E | Static, Full glitch |
| ▖ | U+2596 | Static, Full glitch |
| ▝ | U+259D | Static, Full glitch |
| � | U+FFFD | Corrupted |
� is the Unicode replacement character. Unicode defines it for replacing "an incoming character whose value is unknown or unrepresentable in Unicode", which is why it appears when a file or page is decoded wrongly, and why it reads as corruption.
Binary and Hex show your text the way a computer stores it in UTF-8, the encoding that uses 1 to 4 bytes per character. Plain English letters take one byte. Accents, symbols, and emoji take more.
| Character | Code point | Bytes | Hex | Binary |
|---|---|---|---|---|
| A | U+0041 | 1 | 41 | 01000001 |
| é | U+00E9 | 2 | C3 A9 | 11000011 10101001 |
| € | U+20AC | 3 | E2 82 AC | 11100010 10000010 10101100 |
| 😀 | U+1F600 | 4 | F0 9F 98 80 | 11110000 10011111 10011000 10000000 |
Clean up decodes text made only of 8-digit binary groups or 2-digit hex pairs back into words, emoji included. For more byte tools, use the Binary Code Generator.
Mojibake, Japanese for "character transformation", is the garbled text you get when bytes are decoded with the wrong encoding. The most common case is UTF-8 read as Windows-1252, where each multi-byte character falls apart into two or three symbols.
| Original | UTF-8 bytes | Read as Windows-1252 |
|---|---|---|
| café | 63 61 66 C3 A9 | café |
| naïve | 6E 61 C3 AF 76 65 | naïve |
| über | C3 BC 62 65 72 | über |
| it’s | 69 74 E2 80 99 73 | it’s |
| a—b | 61 E2 80 94 62 | a—b |
| 5 € | 35 20 E2 82 AC | 5 € |
é is stored as the bytes C3 A9. Windows-1252 reads C3 as à and A9 as ©, so "café" turns into "café". With Also repair mojibake ticked, Clean up turns those characters back into bytes and reads them correctly as UTF-8.
Leet, from "elite" and often written 1337, began on bulletin board systems in the 1980s. The Leet style uses the classic digit swaps below. For more letters and symbols, try the Leetspeak Generator.
Choose Clean up in the generator and paste the text. It works in four steps:
Accents typed as one character, like é, survive. Letters replaced with �, flipped capitals, and leet digits cannot be worked out again, so keep a copy of your original text.
| Year | Event |
|---|---|
| 1962 | Astronaut John Glenn explains the word in the book Into Orbit: "Literally, a glitch is a spike or change in voltage in an electrical circuit." Dictionaries trace it to radio slang and possibly the Yiddish glitsh, "slippery place." |
| 1980s | Leet, from "elite", grows on bulletin board systems, where users swap letters for digits and symbols, partly to slip past word filters. |
| 1981 | The original IBM PC uses code page 437, whose bytes B0, B1, and B2 hold the ░ ▒ ▓ shade characters that later become Unicode U+2591 to U+2593. |
| 2004 | On July 27, Something Awful member Shmorky uploads edited comic strips to his website. Two of them mention Zalgo, the source of Zalgo text. |
| 2009 | Zalgo edits reach 4chan in March. In November, Stack Overflow user bobince posts the famous answer on parsing HTML with regex, which dissolves into Zalgo text. |
Making glitch text, the glitch level and reroll, character counts, mojibake, and cleaning glitch text back to normal.
Explore more tools in the directory.
Classic Zalgo text with fine control.
Zalgo, Fullwidth, Chaos font, and Upside-down curses.
Haunted, Dripping, Ritual, and Nightmare horror styles.
Leet with more letter options.
Text to binary and back, with full UTF-8 decoding.
Odd and playful Unicode text styles.