Compression Methods

This document specifies the compression methods supported in 7z archives.

Overview

Compression methods transform data to reduce size. Each method has:

• A unique method ID
• Optional properties configuring the algorithm
• Input/output stream requirements

Method Categories

Category	Description
Copy	No compression
Standard	Methods in official 7-Zip
Extended	Methods in 7-Zip-zstd fork

Support Levels

Level	Meaning
Mandatory	MUST support for compliance
Recommended	SHOULD support for interoperability
Optional	MAY support

Copy (No Compression)

Method ID: 0x00Properties: None Support: Mandatory

Passes data through unchanged. Used for:

• Already compressed data
• Small files where compression overhead exceeds savings
• Testing

LZMA

Method ID: 0x03 0x01 0x01Properties: 5 bytes Support: Mandatory

The original Lempel-Ziv-Markov chain algorithm.

LZMA Properties

Byte	Description
0	Encoder parameters: lc + lp * 9 + pb * 45
1-4	Dictionary size (UINT32, little-endian)

Parameter constraints:

• lc (literal context bits): 0-8
• lp (literal position bits): 0-4
• pb (position bits): 0-4
• lc + lp ≤ 4

Constraint violation: If any parameter exceeds its valid range, or if lc + lp > 4, implementations MUST reject the archive with ERROR_ARCHIVE. These constraints are enforced by the LZMA algorithm; violating them produces undefined decoder behavior.

Default values: lc=3, lp=0, pb=2 → byte 0 = 0x5D

Dictionary size: Power of 2 from 4 KiB to 4 GiB (typical: 16 MiB)

Example Properties

5D 00 00 10 00          # lc=3, lp=0, pb=2, dict=16 MiB (0x01000000)
5D 00 00 00 02          # lc=3, lp=0, pb=2, dict=32 MiB (0x02000000)

LZMA2

Method ID: 0x21Properties: 1 byte Support: Mandatory

Modern LZMA variant with:

• Reset capability for parallel encoding
• Improved handling of incompressible data
• Single-byte properties

LZMA2 Properties

Single byte encoding dictionary size:

Value (decimal)	Value (hex)	Dictionary Size
0-39	0x00-0x27	Calculated (see formula)
40	0x28	4 GiB - 1 (0xFFFFFFFF)
41+	0x29+	Reserved; implementations SHOULD reject

Dictionary size formula (for values 0-39):

base = 2 + (value & 1)
shift = value / 2 + 11
dict_size = base << shift

Common values:

Byte	Dictionary Size
0x14	1 MiB
0x15	1.5 MiB
0x16	2 MiB
0x17	3 MiB
0x18	4 MiB
0x19	6 MiB
0x1A	8 MiB
0x1B	12 MiB
0x1C	16 MiB
0x1D	24 MiB
0x1E	32 MiB
0x1F	48 MiB
0x20	64 MiB

Deflate

Method ID: 0x04 0x01 0x08Properties: None Support: Recommended

ZIP-compatible deflate compression.

• Uses LZ77 + Huffman coding
• Maximum dictionary size: 32 KiB
• Good for compatibility with ZIP tools

Deflate64

Method ID: 0x04 0x01 0x09Properties: None Support: Optional

Enhanced deflate with 64 KiB dictionary. Proprietary Microsoft extension.

• Decompression: SHOULD support
• Compression: MAY support

BZip2

Method ID: 0x04 0x02 0x02Properties: None Support: Recommended

Burrows-Wheeler transform compression.

• Better compression than Deflate
• Slower than LZMA/LZMA2
• Block-based (good for parallel processing)

PPMd

Method ID: 0x03 0x04 0x01Properties: 5 bytes Support: Recommended

Prediction by Partial Matching compression.

PPMd Properties

Bytes	Description
0	Order (model order, typically 2-16)
1-4	Memory size in bytes (UINT32, little-endian)

Typical values:

• Order: 6-8 for text, 2-4 for binary
• Memory: 16 MiB to 256 MiB

Example:

06 00 00 00 10          # Order=6, Memory=256 MiB (0x10000000)

Extended Methods (7-Zip-zstd)

These methods are supported by the 7-Zip-zstd fork and compatible implementations.

Zstandard (Zstd)

Method ID: 0x04 0xF7 0x11 0x01Properties: Variable (1-5 bytes) Support: Optional

Facebook's modern compression algorithm.

Properties format:

Byte	Description
0	Compression level (1-22)
1-4	Dictionary size (optional)

Brotli

Method ID: 0x04 0xF7 0x11 0x02Properties: 1-3 bytes Support: Optional

Google's compression algorithm, optimized for web content.

Properties:

Byte	Description
0	Quality (0-11)
1	Window size (optional)

LZ4

Method ID: 0x04 0xF7 0x11 0x04Properties: Variable Support: Optional

Extremely fast compression with moderate ratio.

Lizard (LZ5)

Method ID: 0x04 0xF7 0x11 0x06Properties: Variable Support: Optional

Balance between LZ4 speed and better compression.

Method Selection Guidance

Use Case	Recommended Method
General purpose	LZMA2
Maximum compression	LZMA2 with large dictionary
Fast compression	Zstd or LZ4
Text files	PPMd
ZIP compatibility	Deflate
Pre-compressed data	Copy
Executables	BCJ + LZMA2

Compression Levels

Most methods support compression levels affecting the speed/ratio trade-off:

Level	Description
0	Store only / fastest
1-3	Fast compression
4-6	Balanced
7-9	Maximum compression

Implementation-specific; not stored in archive.

Stream Requirements

Method	Inputs	Outputs	Notes
Copy	1	1	Simple
LZMA	1	1	Simple
LZMA2	1	1	Simple
Deflate	1	1	Simple
BZip2	1	1	Simple
PPMd	1	1	Simple
Zstd	1	1	Simple

All standard compression methods are simple coders (1 input, 1 output).

Implementation Notes

Decompression

1. Read coder properties from folder definition
2. Initialize decoder with properties
3. Feed pack stream data to decoder
4. Collect unpack stream output
5. Verify output size matches expected unpack size

Compression

1. Select method based on content analysis or user preference
2. Configure encoder with appropriate level/dictionary
3. Compress data to produce pack stream
4. Record method ID and properties in folder definition
5. Record pack size and unpack size

See Also

• Unpack Info - How coders are defined in folders
• Filters - Pre-compression filters
• Encryption - Encryption coder
• Method IDs - Complete method ID table