Tool for hiding a secret message within a text or a tweet using steganography Steg of the Dump by Holloway, which employs Unicode homoglyphs.
Steg of the Dump - dCode
Tag(s) : Steganography
dCode is free and its tools are a valuable help in games, maths, geocaching, puzzles and problems to solve every day!
A suggestion ? a feedback ? a bug ? an idea ? Write to dCode!
Steg of the Dump is a textual steganography tool created by Holloway. It hides a secret message inside a seemingly innocuous public text (called the carrier text). The result, called the stegotext, broadly retains the appearance of the original text, but some characters are replaced with visually similar Unicode characters (homoglyphs).
Steg of the Dump encoding follows several steps:
1. Convert each character of the hidden message into a 6-bit number, corresponding to its position in a 48-symbol alphabet (see below).
2. Scan the text and identify the characters present in the substitution table.
3. For each character, take as many bits as its number of Unicode variants allows. Keep the original character to represent a sequence of zeros; use the variants to represent the other possible values.
4. Copy the unmodified characters and the remaining characters into the result.
Each character with $ n $ replacement variants has $ n+1 $ possible states: the original character and its variants. The number of bits that can be encoded at that position is $ k = \lfloor \log_2(n+1) \rfloor $.
Example: A character with a single variant can encode 1 bit. A character with 3 variants has four states and can therefore encode 2 bits.
Decoding a Steg of the Dump stegotext involves scanning the text and recognising the characters present in the substitution table.
1. Assign a binary sequence to each recognised character: the original character represents zeros, while each variant represents a different binary value.
2. Split the stream into groups of 6 bits and convert each group into a symbol from the hidden-message alphabet. If necessary, pad the binary stream with zeros until its length is a multiple of 6 bits.
The hidden message uses an alphabet of 48 symbols, indexed from 0 to 47. Binary values between 48 and 63 therefore correspond to no symbol and are ignored.
A homoglyph is a character whose appearance is identical or very similar to that of another character, even though their Unicode code points are different.
Example: The Latin letter A (U+0041), the Greek letter Alpha (U+0391) and the Cyrillic letter A (U+0410) can be visually difficult to distinguish, depending on the font used.
In Steg of the Dump, substitutions are defined by a table. Not all pairs of visually similar characters are therefore necessarily usable by the tool.
The substitution table is as follows:
| ! | " | $ | % | & | ' | ( | ) |
| * | + | , | - | . | / | 0 | 1 |
| 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| : | ; | < | = | > | ? | @ | A,Α,А |
| B,Β,В | C,Ϲ,Ⅽ | D | E,Ε,Е | F | G | H,Η,Н | I,Ι,І |
| J | K,Κ,K | L | M,Μ,М | N | O,Ο,О | P,Ρ,Р | Q |
| R | S | T,Τ,Т | U | V,Ѵ,Ⅳ | W | X,Χ,Ⅹ | Y,Υ,Ү |
| Z | [ | \ | ] | ^ | _ | ` | a |
| b | c,ϲ,с | d | e | f | g | h | i,і,ⅰ |
| j | k | l | m | n | o,ο,о | p | q |
| r | s | t | u | v,ν,ⅴ | w | x,х,ⅹ | y |
| z | { | | | } | ~ |
And spaces can be encoded with U+2000, U+2001, U+2002, U+2003, U+2004, U+2005, U+2006, U+2007, U+2008, U+2009, U+200A, U+2028, U+2029, U+202F and U+205F.
The Steg of the Dump table uses several categories of characters:
— Fullwidth forms: the characters in the Unicode block “Halfwidth and Fullwidth Forms” (U+FF01 to U+FF5E) are fullwidth variants of printable ASCII characters. The hash sign (#) is absent.
— Greek and Cyrillic letters: some Latin letters can be replaced by visually similar characters, for example Latin O by Greek omicron (U+039F) or Cyrillic O (U+041E).
— Other confusable characters: some letters have variants such as the Kelvin sign (U+212A), Roman numerals or letters from other alphabets.
— Unicode spaces: the table includes the characters U+2000 to U+200A, U+2028, U+2029, U+202F and U+205F. These 15 variants of the ordinary space can represent up to 4 bits per position.
The number of bits available depends on the number of variants actually listed for each character. The letters A, B, C, E, H, I, K, M, O, P, T, V, X, Y, c, i, o, v and x each have three variants in the table, in addition to the original character, and can therefore encode 2 bits per position.
The hidden message uses an alphabet limited to 48 symbols: abcdefghijklmnopqrstuvwxyz123456789'0.:/\%-_?&;
— The space;
— Lowercase letters from a to z;
— Digits from 0 to 9;
— The following signs: apostrophe, full stop, colon, forward slash, backslash, percent sign, hyphen, underscore, question mark, ampersand and semicolon.
Uppercase letters entered are converted to lowercase. Accented characters and most other punctuation marks are not part of the supported alphabet.
This restriction makes it possible to represent each symbol on 6 bits, since $ 2^6 = 64 $ and the alphabet contains only 48 symbols.
The theoretical capacity, expressed in bits, is calculated by adding the capacity of each carrier character:
$ C = \sum_i \lfloor \log_2(n_i+1) \rfloor $
where $ n_i $ is the number of replacement variants available for the carrier character at position $ i $.
The capacity in hidden symbols is approximately $ \lfloor C/6 \rfloor $, since each symbol uses 6 bits.
Example: The text dCode breaks down as follows: d: 1 bit; C: 2 bits; o: 2 bits; d: 1 bit; e: 1 bit. The total capacity is 7 bits.
A hidden message of $ m $ characters requires at least $ 6(m+1) $ bits.
As a guide, an average capacity of 1 to 2 bits per carrier character corresponds to about 3 to 6 carrier characters per hidden character. Spaces and some uppercase letters are particularly useful, as they have more variants.
The presence of Unicode substitutions, such as those used in Steg of the Dump, gives a message an irregular appearance: certain letters or symbols may exhibit variations in width (fixed-pitch) and spacing (using different types of em spaces/quadrats).
The reference implementation is available on Matthew Holloway’s website (here it was originally designed for hiding messages within tweets (on Twitter).
The acronym SOTD or the name Matthew Holloway serve as clues.
dCode retains ownership of the "Steg of the Dump" source code. Any algorithm for the "Steg of the Dump" algorithm, applet or snippet or script (converter, solver, encryption / decryption, encoding / decoding, ciphering / deciphering, breaker, translator), or any "Steg of the Dump" functions (calculate, convert, solve, decrypt / encrypt, decipher / cipher, decode / encode, translate) written in any informatic language (Python, Java, PHP, C#, Javascript, Matlab, etc.) or any database download or API access for "Steg of the Dump" or any other element are not public (except explicit open source licence). Same with the download for offline use on PC, mobile, tablet, iPhone or Android app.
Reminder: dCode is an educational and teaching resource, accessible online for free and for everyone.
The content of the page "Steg of the Dump" and its results may be freely copied and reused, including for commercial purposes, provided that dCode.fr is cited as the source (Creative Commons CC-BY free distribution license).
Exporting the results is free and can be done simply by clicking on the export icons ⤓ (.csv or .txt format) or ⧉ (copy and paste).
To cite dCode.fr on another website, use the link:
In a scientific article or book, the recommended bibliographic citation is: Steg of the Dump on dCode.fr [online website], retrieved on 2026-11-19,