Joint Photographic Experts Group (.jpg)

ISO/IEC 標準

JPEG is the most widely adopted lossy compression standard for digital photography and web imagery, exploiting the human eye's higher sensitivity to luminance over chrominance.

発明者: Joint Photographic Experts Group年: 1992MIME: image/jpeg

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 →

JPG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

JPG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 JPG 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 JPG ファイルを識別します:

16進数シグネチャ (オフセット 0):

FF D8 FF

ASCII表現: ÿØÿ

標準化: ISO/IEC 10918-1 / ITU-T T.81

技術仕様

コンテナアーキテクチャ	Marker-delimited binary stream (0xFF prefix) wrapped in JFIF or EXIF headers
圧縮	Lossy Discrete Cosine Transform (DCT) + Run-length & Huffman entropy coding
バイトエンディアン	Big-Endian for marker lengths and quantization tables; Exif can be Little (II) or Big (MM)
色空間	sRGB, Adobe RGB, CMYK, Y'CbCr, Grayscale
チャンネルと構造	8-bit per channel (24-bit Truecolor) or 8-bit Grayscale
最大寸法	65,535 x 65,535 pixels (2-byte unsigned integers)
透過性	None (Opaque only; no alpha channel support)
ストリーミングとプログレッシブ	Progressive JPEG encoding stores multiple spectral passes from low to high resolution

技術比較マトリックス: JPG 対競合製品

技術属性	JPG (現在)	WebP	AVIF	PNG
Compression Paradigm	Lossy DCT + Huffman coding	Lossy (VP8) & Lossless	Lossy (AV1) & Lossless	Lossless Deflate
Alpha Transparency	No alpha channel support (always opaque)	8-bit alpha transparency	8-bit / 10-bit / 12-bit alpha	Full 8-bit / 16-bit alpha
Hardware Decoding	100% dedicated silicon in virtually every GPU and mobile SoC	Widespread hardware decode in modern SoCs	Rapidly expanding hardware decode in modern GPUs	CPU-based fast decode
Compression Efficiency	Good baseline for photographs	25%-35% smaller than JPEG at equal quality	40%-55% smaller than JPEG at equal quality	Inefficient for photographic content
Color Bit Depth	Standard 8-bit per channel (24-bit Truecolor)	8-bit per channel	10-bit & 12-bit HDR wide color gamut	8-bit & 16-bit per channel

一般的な破損モードと16進数リカバリガイド

⚠️ Image viewer shows 'Premature end of JPEG file' or the lower portion of the photo is solid gray or colorful banding.

根本原因: Truncated download or incomplete storage write missing the 0xFFD9 (EOI) trailer marker and image scan payload.

復旧: Append marker 0xFFD9 to the end of the truncated stream to allow decoders to display the intact upper portion of the photograph.

⚠️ Image colors appear heavily shifted, inverted, or psychedelic.

根本原因: CMYK color space header inverted or invalid embedded ICC profile.

復旧: Convert the color space from CMYK to sRGB using ImageMagick ('magick broken.jpg -colorspace sRGB fixed.jpg').

⚠️ File fails to open and reports 'Invalid marker found'.

根本原因: Bit corruption inside the Quantization Table (DQT 0xFFDB) or Huffman Table (DHT 0xFFC4) headers.

復旧: Replace the damaged header markers with standard baseline JFIF Huffman and Quantization tables extracted from an intact JPEG.

セキュリティ分析とパーサー攻撃ベクトル

JPEG security primarily involves metadata privacy leaks (EXIF) and buffer overflows in legacy C decoders (libjpeg).

既知の攻撃ベクター

Sensitive metadata disclosure: Unstripped EXIF headers containing GPS coordinates, camera serial numbers, and thumbnail previews.
Buffer over-reads and integer overflows in legacy libjpeg parsing corrupt restart markers (RST) or progressive scan scripts.
Polyglot files combining JPEG headers with executable ZIP or RAR archives.

防御的ベストプラクティス: Always strip EXIF and GPS geolocation metadata when publishing user photos to protect user privacy. Use modern memory-safe decoders (such as image-rs in Rust) or sandboxed WebAssembly decoders.

歴史的背景とマイルストーン

1998JFIF and EXIF container standards establish digital camera dominance.

1992JPEG specification ratified as ISO/IEC 10918-1.

1986Joint Photographic Experts Group founded under ISO and CCITT.

主な利点とメリット

100% universal hardware and software decoding support across all computers and cameras.
Excellent compression ratios (10:1 to 20:1) for photographic continuous tones.
Small payload footprint with progressive rendering capability.

技術的な制限とデメリット

Lossy compression causes compression artifacts (ringing and blockiness) along sharp edges.
No transparency channel support.
Generation loss occurs every time a file is decompressed, edited, and re-saved.

興味深い技術トリビア

The classic 'Lenna' test image used to develop JPEG compression was scanned from the November 1972 issue of Playboy magazine.
JPEG 4:2:0 subsampling halves chroma resolution both horizontally and vertically with near-zero perceptual degradation.
Every standard JPEG begins with marker 0xFFD8 (Start of Image) and ends with 0xFFD9 (End of Image).

よくある技術的な質問

Why does saving a JPEG file multiple times reduce image quality?

JPEG is an irreversibly lossy compression format. Each time you decompress and re-save a JPEG, the DCT quantization step discards high-frequency spatial details and re-approximates 8x8 pixel blocks, causing cumulative 'generation loss' artifacts.

What is Progressive JPEG and why is it useful?

A standard baseline JPEG decodes scanline by scanline from top to bottom. Progressive JPEG encodes image data in multiple spectral passes, allowing browsers to render a low-resolution blur of the entire image instantly and sharpen it progressively as bytes download.

What is Chroma Subsampling (4:2:0 vs 4:4:4)?

The human retina has more rod cells (luminance/brightness) than cone cells (color). JPEG exploits this by preserving full brightness resolution (Y) while halving color resolution (Cb, Cr) horizontally and vertically (4:2:0), shedding 50% of the raw data before compression with imperceptible loss.

Can a JPEG image have a transparent background?

No. The standard JPEG specification has no concept of an alpha channel. If transparency is required, you must convert the image to WebP, AVIF, or PNG.