colopresso 12.3.3

PNG conversion and compression library

Colopresso Python Bindings

Python bindings for the colopresso image compression library. Encode PNG images to WebP, AVIF, or optimized PNG (PNGX) formats.

Table of Contents

• Requirements
• Installation
• Quick Start
• API Reference
  • Encoding Functions
  • Config Class
  • PngxLossyType Enum
  • Configuration Parameters
  • Utility Functions
  • Exception Classes
• Examples
• Building Wheels

Requirements

[!IMPORTANT] AVX2 instruction set support is required on x86_64 (amd64) platforms. Intel Haswell (2013) or later, or AMD Excavator or later processors are required.

Platform	Architecture	Requirements
Windows	x64	AVX2 support
Windows	ARM64	NEON (ASIMD)
macOS	x86_64	AVX2 support
macOS	arm64 (Apple Silicon)	NEON (ASIMD)
Linux (glibc)	x86_64	AVX2 support
Linux (glibc)	aarch64	NEON (ASIMD)
Linux (musl)	x86_64	AVX2 support
Linux (musl)	aarch64	NEON (ASIMD)

Installation

From PyPI

pip install "colopresso"

From source (requires CMake, Rust, and C compiler)

cd python
pip install .

Development installation

cd python
pip install -e ".[dev]"

Quick Start

import colopresso

# Read PNG file
with open("input.png", "rb") as f:
    png_data = f.read()

# Convert to WebP
webp_data = colopresso.encode_webp(png_data)
with open("output.webp", "wb") as f:
    f.write(webp_data)

# Convert to AVIF
avif_data = colopresso.encode_avif(png_data)
with open("output.avif", "wb") as f:
    f.write(avif_data)

# Optimize PNG
optimized_png = colopresso.encode_pngx(png_data)
with open("output.png", "wb") as f:
    f.write(optimized_png)

---

API Reference

Encoding Functions

encode_webp(png_data, config=None) -> bytes

Encode PNG data to WebP format.

Parameters:

• png_data (bytes): Raw PNG file data
• config (Config, optional): Encoding configuration. Uses defaults if not provided.

Returns:

• bytes: WebP encoded data

Raises:

• ColopressoError: If encoding fails

Example:

import colopresso

with open("input.png", "rb") as f:
    png_data = f.read()

# Default settings
webp_data = colopresso.encode_webp(png_data)

# Custom quality
config = colopresso.Config(webp_quality=90.0)
webp_data = colopresso.encode_webp(png_data, config)

---

encode_avif(png_data, config=None) -> bytes

Encode PNG data to AVIF format.

Parameters:

• png_data (bytes): Raw PNG file data
• config (Config, optional): Encoding configuration. Uses defaults if not provided.

Returns:

• bytes: AVIF encoded data

Raises:

• ColopressoError: If encoding fails (including when output is larger than input)

Example:

import colopresso

with open("input.png", "rb") as f:
    png_data = f.read()

config = colopresso.Config(avif_quality=70.0, avif_speed=4)
avif_data = colopresso.encode_avif(png_data, config)

---

encode_pngx(png_data, config=None) -> bytes

Optimize PNG data using the PNGX encoder. Supports both lossless and lossy compression.

Parameters:

• png_data (bytes): Raw PNG file data
• config (Config, optional): Encoding configuration. Uses defaults if not provided.

Returns:

• bytes: Optimized PNG data

Raises:

• ColopressoError: If encoding fails

Example:

import colopresso

with open("input.png", "rb") as f:
    png_data = f.read()

# 256-color palette optimization
config = colopresso.Config(
    pngx_lossy_enable=True,
    pngx_lossy_type=colopresso.PngxLossyType.PALETTE256,
    pngx_lossy_max_colors=256
)
optimized = colopresso.encode_pngx(png_data, config)

---

Config Class

The Config dataclass holds all encoder settings.

@dataclass
class Config:
    # WebP settings
    webp_quality: float = 80.0
    webp_lossless: bool = False
    webp_method: int = 6
    
    # AVIF settings
    avif_quality: float = 50.0
    avif_alpha_quality: int = 100
    avif_lossless: bool = False
    avif_speed: int = 6
    avif_threads: int = 1
    
    # PNGX settings
    pngx_level: int = 5
    pngx_strip_safe: bool = True
    pngx_optimize_alpha: bool = True
    pngx_lossy_enable: bool = True
    pngx_lossy_type: PngxLossyType = PngxLossyType.PALETTE256
    pngx_lossy_max_colors: int = 256
    pngx_lossy_quality_min: int = 80
    pngx_lossy_quality_max: int = 95
    pngx_lossy_speed: int = 3
    pngx_lossy_dither_level: float = 0.6
    pngx_threads: int = 1

---

PngxLossyType Enum

Specifies the type of lossy compression for PNGX.

class PngxLossyType(IntEnum):
    PALETTE256 = 0          # 256-color indexed palette
    LIMITED_RGBA4444 = 1    # RGBA4444 limited (16-bit color)
    REDUCED_RGBA32 = 2      # Reduced color RGBA32

Value	Description	Use Case
PALETTE256	Convert to 256-color indexed palette	Icons, illustrations with few colors
LIMITED_RGBA4444	Limit each RGBA channel to 4 bits	Prevents banding artifacts in RGBA16bit and RGBA4444
REDUCED_RGBA32	Reduce colors while keeping RGBA32	When PALETTE256 is not acceptable

---

Configuration Parameters

WebP Settings

Parameter	Type	Default	Description
webp_quality	float	80.0	Quality (0.0-100.0). Higher = better quality, larger size
webp_lossless	bool	False	True: lossless compression, False: lossy compression
webp_method	int	6	Compression method (0-6). Higher = better compression, slower

Advanced WebP Parameters:

Parameter	Type	Default	Description
webp_target_size	int	0	Target file size in bytes. 0 = disabled
webp_target_psnr	float	0.0	Target PSNR in dB. 0 = disabled
webp_segments	int	4	Number of segments (1-4)
webp_sns_strength	int	50	Spatial noise shaping strength (0-100)
webp_filter_strength	int	60	Deblocking filter strength (0-100)
webp_filter_sharpness	int	0	Filter sharpness (0-7). 0 = sharpest
webp_filter_type	int	1	Filter type. 0 = simple, 1 = strong
webp_autofilter	bool	True	Auto-adjust filter strength
webp_alpha_compression	bool	True	Compress alpha channel
webp_alpha_filtering	int	1	Alpha filtering (0-2)
webp_alpha_quality	int	100	Alpha channel quality (0-100)
webp_pass	int	1	Number of encoding passes (1-10)
webp_preprocessing	int	0	Preprocessing. 0=none, 1=segment-smooth, 2=pseudo-random dither
webp_partitions	int	0	Number of partitions (0-3)
webp_partition_limit	int	0	Partition size limit (0-100)
webp_emulate_jpeg_size	bool	False	Emulate JPEG output size
webp_thread_level	int	0	Thread level (0-1)
webp_low_memory	bool	False	Low memory mode
webp_near_lossless	int	100	Near-lossless preprocessing (0-100). 100 = off
webp_exact	bool	False	Preserve RGB values in transparent areas
webp_use_delta_palette	bool	False	Use delta palette (experimental)
webp_use_sharp_yuv	bool	False	Use sharp RGB to YUV conversion

---

AVIF Settings

Parameter	Type	Default	Description
avif_quality	float	50.0	Quality (0.0-100.0). Higher = better quality, larger size
avif_alpha_quality	int	100	Alpha channel quality (0-100)
avif_lossless	bool	False	True: lossless compression, False: lossy compression
avif_speed	int	6	Encoding speed (0-10). 0 = best quality/slowest, 10 = lowest quality/fastest
avif_threads	int	1	Number of threads for encoding

Recommended Settings:

Use Case	quality	speed	Description
High Quality	80-90	4	High quality web images
Balanced	50-60	6	General purpose
Fast	40-50	8-10	Batch processing, previews

---

PNGX Settings

Basic Settings

Parameter	Type	Default	Description
pngx_level	int	5	Compression level (1-6). Higher = better compression
pngx_strip_safe	bool	True	Remove safely removable metadata
pngx_optimize_alpha	bool	True	Optimize color info in transparent pixels
pngx_threads	int	1	Number of threads for processing

Lossy Compression Settings

Parameter	Type	Default	Description
pngx_lossy_enable	bool	True	Enable lossy compression
pngx_lossy_type	PngxLossyType	PALETTE256	Type of lossy compression
pngx_lossy_max_colors	int	256	Maximum colors for palette mode (2-256)
pngx_lossy_quality_min	int	80	Minimum quality (0-100)
pngx_lossy_quality_max	int	95	Maximum quality (0-100)
pngx_lossy_speed	int	3	Quantization speed (1-10). Higher = faster, lower quality
pngx_lossy_dither_level	float	0.6	Dithering strength (0.0-1.0)

PALETTE256 Mode Settings

Parameter	Type	Default	Description
pngx_lossy_reduced_colors	int	-1	Target color count after reduction. -1 = auto
pngx_saliency_map_enable	bool	True	Adaptive quantization using saliency map
pngx_chroma_anchor_enable	bool	True	Preserve chroma anchor points
pngx_adaptive_dither_enable	bool	True	Adaptive dithering
pngx_gradient_boost_enable	bool	True	Gradient enhancement
pngx_chroma_weight_enable	bool	True	Apply chroma weighting
pngx_postprocess_smooth_enable	bool	True	Post-processing smoothing
pngx_postprocess_smooth_importance_cutoff	float	0.6	Smoothing importance cutoff
pngx_palette256_gradient_profile_enable	bool	True	Enable gradient profile
pngx_palette256_gradient_dither_floor	float	0.78	Gradient dither floor
pngx_palette256_alpha_bleed_enable	bool	False	Enable alpha bleed
pngx_palette256_alpha_bleed_max_distance	int	64	Maximum bleed distance in pixels
pngx_palette256_alpha_bleed_opaque_threshold	int	248	Opaque threshold
pngx_palette256_alpha_bleed_soft_limit	int	160	Soft bleed limit

LIMITED_RGBA4444 Mode Settings

Parameter	Type	Default	Description
pngx_lossy_reduced_bits_rgb	int	4	RGB channel bit depth (1-8)
pngx_lossy_reduced_alpha_bits	int	4	Alpha channel bit depth (1-8)

---

Utility Functions

Version Information

def get_version() -> int

Get the colopresso version number.

def get_libwebp_version() -> int

Get the libwebp version number (e.g., 67072 = 1.4.0).

def get_libpng_version() -> int

Get the libpng version number.

def get_libavif_version() -> int

Get the libavif version number.

def get_pngx_oxipng_version() -> int

Get the oxipng version number.

def get_pngx_libimagequant_version() -> int

Get the libimagequant version number.

Build Information

def get_buildtime() -> int

Get the build timestamp (Unix time).

def get_compiler_version_string() -> str

Get the C compiler version string.

def get_rust_version_string() -> str

Get the Rust compiler version string.

Thread Information

def is_threads_enabled() -> bool

Check if multithreading is enabled.

def get_default_thread_count() -> int

Get the default number of threads.

def get_max_thread_count() -> int

Get the maximum available thread count.

---

Exception Classes

ColopressoError

class ColopressoError(Exception):
    code: int       # Error code
    message: str    # Error message

Error Codes:

Code	Name	Description
0	OK	Success
1	File not found	File not found
2	Invalid PNG	Invalid PNG data
3	Invalid format	Invalid format
4	Out of memory	Memory allocation failed
5	Encode failed	Encoding failed
6	Decode failed	Decoding failed
7	IO error	Input/output error
8	Invalid parameter	Invalid parameter
9	Output not smaller	Output is not smaller than input

Example:

import colopresso

try:
    result = colopresso.encode_avif(png_data)
except colopresso.ColopressoError as e:
    if e.code == 9:
        print("AVIF compression was not effective. Using original PNG.")
    else:
        print(f"Error: {e.message}")

---

Examples

WebP Conversion with Quality Settings

import colopresso

with open("photo.png", "rb") as f:
    png_data = f.read()

# High quality WebP
config = colopresso.Config(
    webp_quality=95.0,
    webp_method=6,
    webp_use_sharp_yuv=True
)
high_quality = colopresso.encode_webp(png_data, config)

# High compression WebP
config = colopresso.Config(
    webp_quality=60.0,
    webp_method=6
)
compressed = colopresso.encode_webp(png_data, config)

# Lossless WebP
config = colopresso.Config(webp_lossless=True)
lossless = colopresso.encode_webp(png_data, config)

AVIF with Multithreading

import colopresso

with open("large_image.png", "rb") as f:
    png_data = f.read()

# Multithreaded high-speed encoding
config = colopresso.Config(
    avif_quality=70.0,
    avif_speed=6,
    avif_threads=4
)
avif_data = colopresso.encode_avif(png_data, config)

PNG Optimization (256-Color Palette)

import colopresso

with open("icon.png", "rb") as f:
    png_data = f.read()

config = colopresso.Config(
    pngx_lossy_enable=True,
    pngx_lossy_type=colopresso.PngxLossyType.PALETTE256,
    pngx_lossy_max_colors=256,
    pngx_lossy_quality_min=85,
    pngx_lossy_quality_max=100,
    pngx_lossy_dither_level=0.8
)
optimized = colopresso.encode_pngx(png_data, config)

print(f"Original: {len(png_data)} bytes")
print(f"Optimized: {len(optimized)} bytes")
print(f"Reduction: {(1 - len(optimized)/len(png_data))*100:.1f}%")

PNG Lossless Optimization

import colopresso

with open("screenshot.png", "rb") as f:
    png_data = f.read()

config = colopresso.Config(
    pngx_lossy_enable=False,  # Disable lossy compression
    pngx_level=6,             # Maximum compression level
    pngx_strip_safe=True,     # Remove metadata
    pngx_optimize_alpha=True  # Optimize alpha
)
optimized = colopresso.encode_pngx(png_data, config)

Game Texture PNG (RGBA4444)

import colopresso

with open("texture.png", "rb") as f:
    png_data = f.read()

# RGBA4444 format (16-bit color)
config = colopresso.Config(
    pngx_lossy_enable=True,
    pngx_lossy_type=colopresso.PngxLossyType.LIMITED_RGBA4444,
    pngx_lossy_reduced_bits_rgb=4,
    pngx_lossy_reduced_alpha_bits=4
)
optimized = colopresso.encode_pngx(png_data, config)

Batch Processing

import colopresso
from pathlib import Path
from concurrent.futures import ThreadPoolExecutor

def convert_to_webp(png_path: Path) -> tuple[Path, int, int]:
    """Convert PNG to WebP and return size info"""
    with open(png_path, "rb") as f:
        png_data = f.read()
    
    config = colopresso.Config(webp_quality=80.0)
    webp_data = colopresso.encode_webp(png_data, config)
    
    webp_path = png_path.with_suffix(".webp")
    with open(webp_path, "wb") as f:
        f.write(webp_data)
    
    return webp_path, len(png_data), len(webp_data)

# Parallel conversion
png_files = list(Path("images").glob("*.png"))
with ThreadPoolExecutor(max_workers=4) as executor:
    results = list(executor.map(convert_to_webp, png_files))

for path, original, converted in results:
    print(f"{path.name}: {original} -> {converted} ({(1-converted/original)*100:.1f}% reduction)")

Error Handling

import colopresso

def safe_encode(png_data: bytes, format: str = "webp") -> bytes | None:
    """Encode with error handling"""
    config = colopresso.Config()
    
    try:
        if format == "webp":
            return colopresso.encode_webp(png_data, config)
        elif format == "avif":
            return colopresso.encode_avif(png_data, config)
        elif format == "pngx":
            return colopresso.encode_pngx(png_data, config)
        else:
            raise ValueError(f"Unknown format: {format}")
            
    except colopresso.ColopressoError as e:
        if e.code == 2:
            print("Error: Input is not a valid PNG")
        elif e.code == 9:
            print(f"{format.upper()} compression was not effective")
            return png_data  # Return original data
        else:
            print(f"Encoding error: {e.message}")
        return None

# Usage
with open("input.png", "rb") as f:
    png_data = f.read()

result = safe_encode(png_data, "avif")
if result:
    with open("output.avif", "wb") as f:
        f.write(result)

---

Building Wheels

cd python
pip install build
python -m build --wheel

The wheel will be created in python/dist/.

---

License

colopresso is licensed under GPL-3.0-or-later.

Copyright (C) 2026 COLOPL, Inc.