Portable Network Graphics (.png)

W3C 標準

Portable Network Graphics (PNG) is an unencumbered raster graphics file format that supports lossless data compression, indexed colors, and variable alpha channel transparency.

発明者: PNG Development Group (Thomas Boutell, Glenn Randers-Pehrson)年: 1995MIME: image/png

PNG を JPG に変換

無料のブラウザ内 PNG から JPG へのコンバーター。お使いのデバイスでファイルを即座に変換します。

PNG から JPG へのコンバーター →

PNGをオンラインで圧縮

ブラウザ内100%ローカル圧縮 — ファイルが外部サーバーに送信されることはありません

PNGを圧縮 →

調査とメタデータ

オペレーティングシステムやファイル解析ツールは、先頭のバイナリバイトシーケンスを検査することで PNG ファイルを識別します:

ファイルを選択またはドラッグ

ここにファイルを選択またはドロップ

100%プライベートなブラウザ内変換 - ファイルがデバイスから外部に出ることはありません

または貼り付け Ctrl+V

ネットワーク送信ゼロのプライバシー保証： 外部サーバーへのデータ送信は0バイトです。すべての処理はお使いのブラウザのサンドボックス内でローカルに実行されます。

バイトレベルのヘッダーシグネチャ (マジックバイト)

オペレーティングシステムやファイル解析ツールは、先頭のバイナリバイトシーケンスを検査することで PNG ファイルを識別します:

16進数シグネチャ (オフセット 0):

89 50 4E 47 0D 0A 1A 0A

ASCII表現: .PNG....

標準化: W3C Recommendation (1996) / ISO/IEC 15948:2004

技術仕様

コンテナアーキテクチャ	Chunk-based binary container with 4-byte CRC32 verification per chunk
圧縮	Deflate (LZ77 + Huffman coding) with 5 predictive scanline filters (None, Sub, Up, Average, Paeth)
バイトエンディアン	Big-Endian (network byte order) for chunk lengths, dimensions, and CRCs
色空間	sRGB, Display P3, Indexed Palette (1-8 bit), Grayscale, Truecolor (24/48-bit)
チャンネルと構造	RGB + Alpha (Up to 16 bits per channel / 64-bit RGBA)
最大寸法	2,147,483,647 x 2,147,483,647 pixels (4-byte unsigned integers)
透過性	Full 8-bit / 16-bit variable alpha transparency channel
ストリーミングとプログレッシブ	Adam7 7-pass interlacing algorithm enables progressive 2D scanline rendering

技術比較マトリックス: PNG 対競合製品

技術属性	PNG (現在)	WebP	AVIF	JPEG
Compression Type	Strictly lossless (Deflate + 2D predictive filters)	Lossy (VP8) & Lossless (VP8L)	Lossy (AV1 intra) & Lossless	Strictly lossy (Discrete Cosine Transform)
Alpha Transparency	Full 8-bit / 16-bit variable alpha channel	8-bit alpha channel supported	8-bit / 10-bit / 12-bit alpha supported	No transparency channel support
File Size (Photographs)	Large (typically 3x to 6x larger than lossy formats)	Small (25%-35% smaller than JPEG)	Ultra-small (40%-55% smaller than JPEG)	Medium baseline
File Size (Graphics/UI)	Extremely compact for screenshots, line art, and text	Excellent lossless mode	Good, but slower to encode	Poor (produces ringing artifacts along sharp edges)
Browser Compatibility	100% universal across all browsers since 1998	97%+ modern browser support	93%+ modern browser support	100% universal

一般的な破損モードと16進数リカバリガイド

⚠️ Image viewer displays 'Fatal error: Not a valid PNG file' or fails to open.

根本原因: The first 8 bytes were altered by an ASCII transfer mode (mangling 0x0D 0x0A into 0x0A or 0x0D 0x0D 0x0A).

復旧: Open the file in a hex editor and verify the first 8 bytes match '89 50 4E 47 0D 0A 1A 0A'. Restoring these exact 8 header bytes frequently recovers the image.

⚠️ pngcheck reports 'CRC error in chunk IDAT' or image loads partially with corrupt bottom scanlines.

根本原因: Bit corruption in the compressed zlib datastream or invalid CRC32 checksum in an IDAT chunk.

復旧: Use pngcheck -f or Deflate stream recovery utilities to recalculate valid CRC32 values, or decompress surviving scanlines up to the corrupt block.

⚠️ Image renders fine in desktop viewers but fails in iOS applications.

根本原因: iOS expects Apple's proprietary CgBI byte swapped format, or non-standard color space chunks.

復旧: Normalize the image using 'pngcrush -revert-iphone-optimizations' to restore standard W3C RGBA ordering.

セキュリティ分析とパーサー攻撃ベクトル

PNG is widely parsed across all platforms. Modern security concerns center on parser memory exhaustion and ancillary chunk injection.

既知の攻撃ベクター

Oversized tEXt and zTXt metadata chunks triggering Out-Of-Memory (OOM) denial-of-service in un-sandboxed decoders.
Heap buffer overflow vulnerabilities in legacy libpng releases (e.g. CVE-2015-8126 in png_set_PLTE).
Pixel flood decompression attacks where an image with small file size declares enormous dimensions (e.g. 100,000 x 100,000 pixels).
Steganographic exfiltration hiding encrypted payloads in ancillary private chunks (e.g. eXIf or custom chunk types).

防御的ベストプラクティス: Always validate image dimensions before allocating pixel buffers in memory. Strip non-essential ancillary chunks (eXIf, tEXt) on user uploads using a memory-safe sanitizer.

歴史的背景とマイルストーン

2004Standardized internationally as ISO/IEC 15948.

1996PNG 1.0 published as a W3C Recommendation.

1995Created on Usenet comp.graphics as an open replacement for GIF following Unisys's LZW patent fee enforcement.

主な利点とメリット

Lossless compression guarantees zero pixel degradation or generation loss upon saving.
Smooth 8-bit alpha transparency enables clean blending on complex backgrounds.
Gamma correction and chromaticity chunks ensure consistent rendering across diverse monitors.

技術的な制限とデメリット

Significantly larger file sizes compared to modern lossy formats like WebP or AVIF for photographic content.
No native multi-frame animation in standard PNG (requires APNG extension).
High bit-depth (16-bit per channel) creates huge uncompressed in-memory decode buffers.

興味深い技術トリビア

The 8-byte magic header was specifically designed to detect CR/LF line ending corruption by FTP transfers.
The unofficial recursive acronym for PNG was 'PNG's Not GIF'.
Apple uses a proprietary CgBI chunk in iOS app PNGs that reverses RGBA to BGRA for direct GPU rendering.

よくある技術的な質問

Why does a PNG file size sometimes increase when saved in image editors?

Different software suites use varying Deflate compression levels and predictive scanline filter heuristics. Programs like Photoshop frequently append uncompressed color profiles, XMP metadata blocks, and unoptimized IDAT chunk sizes. Dedicated tools like File2File Compressor re-evaluate all 5 scanline filters to achieve minimal file size.

What is the difference between PNG-8 and PNG-24?

PNG-8 uses an indexed color palette of up to 256 colors (similar to GIF), producing very small files suitable for simple icons. PNG-24 supports 16.7 million colors (Truecolor 8 bits per channel), with PNG-32 adding an 8-bit alpha transparency channel.

Why does PNG have 5 predictive scanline filters?

Deflate compresses repeated byte patterns. Because photographic or gradient pixel values change gradually, PNG filters each scanline (None, Sub, Up, Average, Paeth) by calculating differences between adjacent pixels. This transforms gradient data into strings of near-zero bytes that compress dramatically better under Deflate.

Does PNG support CMYK color spaces for commercial printing?

No. The official W3C PNG specification deliberately restricts color types to Grayscale, Indexed, and RGB. Professional print workflows requiring CMYK must use TIFF or PDF containers.