About

Base45 encoding (defined in RFC 9285) was designed for QR-code-optimized data transport, notably used in EU Digital COVID Certificates. It encodes arbitrary binary data using 45 characters from the alphanumeric QR code subset, achieving roughly 2 input bytes per 3 output characters. Base64, by contrast, uses 64 characters and encodes 3 input bytes into 4 output characters. Converting between these two formats requires an intermediate binary step: decode one encoding to its raw byte sequence, then re-encode into the target format. Mishandling the alphabet mapping or chunk boundaries produces corrupted output with no obvious error indicator. This tool performs the full decode - re-encode pipeline with strict validation against both alphabets.

Limitations: this converter operates on the raw encoded payload. It does not interpret compressed content (e.g., ZLIB-wrapped CBOR inside health certificates). If your Base45 payload contains compressed data, the Base64 output represents those compressed bytes, not the decompressed content. Pro tip: validate your Base45 input length. A valid Base45 string has length L where L mod 3 must equal 0 or 2. A remainder of 1 signals a malformed input.

Formulas

Base45 encoding maps groups of bytes to characters using base-45 positional notation. The alphabet A is indexed 0 - 44.

Encoding 2 bytes [b₀, b₁] to 3 characters:

n = b₀ × 256 + b₁
c₀ = n mod 45
c₁ = (n ÷ 45) mod 45
c₂ = (n ÷ 45²) mod 45

Decoding 3 characters [c₀, c₁, c₂] back to 2 bytes:

n = c₀ + c₁ × 45 + c₂ × 45²
b₀ = (n ÷ 256) mod 256
b₁ = n mod 256

Single trailing byte [b₀] encodes to 2 characters:

c₀ = b₀ mod 45
c₁ = (b₀ ÷ 45) mod 45

where b_i = byte value (0 - 255), c_i = index into the Base45 alphabet, n = intermediate integer, and ÷ denotes integer (floor) division. Overflow check: for a 3-character group, n must satisfy n ≤ 65535 (0xFFFF). For Base64, the standard 6-bit packing is used via native btoa/atob with a Uint8Array binary bridge.

Reference Data

Property	Base45 (RFC 9285)	Base64 (RFC 4648)
Alphabet Size	45 characters	64 characters + pad
Character Set	0-9 A - Z SPACE $ % * + - . / :	A - Z a - z 0-9 + /
Padding Character	None	=
Encoding Ratio	2 bytes → 3 chars (1.5×)	3 bytes → 4 chars (1.33×)
Efficiency	66.7%	75%
QR Alphanumeric Mode	Fully compatible	Requires binary mode
Case Sensitivity	Uppercase only	Case-sensitive
Specification	RFC 9285 (2022)	RFC 4648 (2006)
Primary Use Case	QR codes, health certificates	Email, data URIs, web APIs
Whitespace Handling	Space (0x20) is valid char (index 36)	Whitespace ignored by liberal parsers
Chunk Input Size	2 bytes (last chunk: 1 byte)	3 bytes (padded)
Chunk Output Size	3 chars (last chunk: 2 chars)	4 chars (with = padding)
Max Value per Chunk	2-byte chunk: 45³ − 1 = 91124 (> 65535)	N/A (bit-packing)
1-byte Chunk Max	45² − 1 = 2024 (> 255)	N/A
URL Safe Variant	Not defined	RFC 4648 §5 (- _ replace + /)
Error Detection	Invalid chars or overflow	Invalid chars or bad padding
Sorting Preservation	No	Yes (lexicographic)
Compression Friendly	Often wraps ZLIB+CBOR	Not inherently

Frequently Asked Questions

A 3-character Base45 group decodes to an integer n = c₀ + c₁ × 45 + c₂ × 45². This value must fit in 2 bytes, meaning n ≤ 65535. If the three characters produce a value exceeding this limit, the input is malformed. Check for corrupted or truncated data.

Yes. The converter treats the payload as opaque binary data. It decodes Base64 to raw bytes using atob, then re-encodes those exact bytes into Base45. No interpretation of the content occurs. Compressed payloads (ZLIB, gzip) remain compressed in the output.

Base45 has lower encoding efficiency. It uses 3 characters per 2 bytes (expansion factor 1.5×), while Base64 uses 4 characters per 3 bytes (factor 1.33×). The tradeoff is QR code compatibility: Base45 characters all belong to the QR alphanumeric subset, yielding smaller QR codes despite longer strings.

The space character (ASCII 0x20) is the 37th character in the Base45 alphabet (index 36). It is a valid encoding character, not whitespace to be trimmed. The converter preserves spaces within the input. Leading and trailing spaces are also treated as valid Base45 data.

A valid Base45 string length L satisfies L mod 3 ∈ {0, 2}. A remainder of 1 is structurally impossible under RFC 9285 because a single character cannot represent any valid byte group. The converter rejects such inputs with a specific error message.

Yes. The converter auto-detects Base64url input by checking for - and _ characters (which replace + and / per RFC 4648 §5). It normalizes them to standard Base64 before decoding. Output is always standard Base64 with + and /.