Lossy or lossless: which compression should I use for web images?

Use lossy (JPG, WebP, AVIF) for photographs — the human eye barely notices the difference at quality 80–85, and file sizes are 5–20× smaller. Use lossless (PNG, WebP-lossless) for screenshots, UI mockups, logos and anything with sharp edges or text — lossy creates ugly artifacts around boundaries.

Does conversion preserve EXIF metadata (camera, GPS, date)?

It depends on the tool: ImageMagick and FFmpeg copy EXIF by default, while cwebp drops it unless you pass -metadata all . KaijuConverter strips metadata server-side after conversion to protect your privacy — if you need to keep camera or GPS data, use a desktop tool you control.

Lossy or lossless: which compression should I use for web images?

Use lossy (JPG, WebP, AVIF) for photographs — the human eye barely notices the difference at quality 80–85, and file sizes are 5–20× smaller. Use lossless (PNG, WebP-lossless) for screenshots, UI mockups, logos and anything with sharp edges or text — lossy creates ugly artifacts around boundaries.

Does conversion preserve EXIF metadata (camera, GPS, date)?

It depends on the tool: ImageMagick and FFmpeg copy EXIF by default, while cwebp drops it unless you pass -metadata all . KaijuConverter strips metadata server-side after conversion to protect your privacy — if you need to keep camera or GPS data, use a desktop tool you control.

Guide

JPEG XL (JXL): The Next-Generation Image Standard That Does Everything

PC By Pablo Cirre • Updated May 7, 2026

JPEG XL (JXL): The Next-Generation Image Standard That Does Everything

JPEG XL (JXL) is an open-source image format developed by Google and Cloudinary, standardized as ISO/IEC 18181 in 2022. It is arguably the most technically comprehensive modern image format: it matches AVIF's compression at high quality, supports lossless recompression of existing JPEGs (enabling lossless JPEG transcoding), offers exceptionally fast decoding, supports 1-bit to 32-bit HDR, and handles animations. Despite its technical excellence, JPEG XL faces a challenging adoption situation: Chrome removed support in 2022 (later reversing course), and Safari is the only major browser with native support.

Core Design Goals

JPEG XL was designed with four explicit goals:

Better than existing formats across all dimensions: Outperform JPEG, PNG, WebP, and GIF in all use cases
Lossless JPEG transcoding: Convert existing JPEG files to JXL losslessly (and back), enabling format migration without quality loss
Practical for image professionals: Supports color profiles, HDR, 1-bit to 32-bit depths, CMYK, multi-frame, and metadata
Deployable everywhere: Fast encode/decode, progressive rendering, decent file sizes at high quality

Compression Technology

JPEG XL uses two completely different compression methods depending on the mode:

Lossy Mode (VarDCT)

For lossy compression, JXL uses VarDCT (Variable DCT Block Sizes):

Adaptive quantization: Unlike JPEG's fixed 8×8 block DCT, VarDCT uses variable block sizes (from 2×2 up to 256×256) selected per-region. Smooth areas use large blocks; detail areas use small blocks.
XYB color space: A custom color space derived from the human visual system, where X+Y, B, and a modified luminance channel correspond to the three cone types of human vision. More perceptually uniform than YCbCr.
Butteraugli perceptual metric: JXL's encoder uses a psychovisual model during encoding to optimally allocate bits — spending more bits where the human eye is most sensitive.
SIMD-optimized decoder: The JXL reference implementation (libjxl) uses AVX2/SSE4.2 SIMD instructions for very fast decoding — typically 2–4× faster than JPEG for equivalent quality.

Lossless Mode (Modular)

For lossless compression, JXL uses the Modular compression scheme:

Prediction-based: Each channel (R, G, B, A) is predicted from previously decoded values using a set of predictors
MA (Meta-Adaptive) trees: Decision trees that adaptively select the best predictor for each region of the image
ANS entropy coding: Asymmetric Numeral Systems for efficient entropy coding of prediction residuals

Lossless JXL is typically 35% smaller than PNG for photographic content and competitive with WebP lossless.

JPEG Lossless Transcoding

The most unique JXL feature: lossless transcoding of existing JPEG files. JXL can store the original JPEG DCT coefficients plus a small overhead for JPEG metadata, resulting in:

No quality loss (mathematically lossless roundtrip JPEG → JXL → JPEG)
Typically 20% smaller than the original JPEG
Enables migrating existing JPEG archives to JXL without quality degradation

# Losslessly transcode JPEG to JXL (preserves original JPEG recoverable)
cjxl --lossless_jpeg=1 input.jpg output.jxl

# Recover original JPEG from lossless JXL transcode
djxl output.jxl recovered.jpg

Progressive Decoding

JXL supports progressive decoding in a more sophisticated way than JPEG's progressive scan:

DC pass: Shows a very low-resolution preview almost immediately
AC passes: Incrementally adds detail in multiple refinement passes
The image becomes usable at 1/8 or 1/16 of full data receipt — useful for slow connections where AVIF shows nothing until fully downloaded

Color Depth and HDR Support

Feature	JXL	AVIF	WebP	JPEG
Max bit depth	32-bit float	12-bit	8-bit	8-bit
HDR	✅ Full	✅ Full	❌	❌
Wide color gamut	✅	✅	❌	❌
CMYK	✅	❌	❌	✅
Arbitrary channels	✅	❌	❌	❌
Scientific imaging	✅	❌	❌	❌

The 32-bit float support and arbitrary channel count make JXL unique among image formats — it can store scientific data, medical imaging, and multi-spectral data that no other standard image format handles.

File Size Comparison

Compared to JPEG at equivalent visual quality (SSIM):

Format	Size vs. JPEG	Encoding Speed	Decoding Speed
JPEG XL	-60%	Medium	Very Fast
AVIF	-50%	Very Slow	Medium
WebP	-30%	Fast	Fast
HEIC	-50%	Slow	Fast (HW)

JXL achieves the best compression ratio of any modern image format at equivalent quality, while maintaining fast decoding speeds.

Encoding JXL

cjxl (libjxl reference encoder)

# Lossy encoding (quality 0=worst, 100=best, default 90)
cjxl --quality 80 input.jpg output.jxl

# Match AVIF quality (quality ~70 JXL ≈ quality 70 AVIF)
cjxl --quality 70 input.jpg output.jxl

# Lossless encoding
cjxl --lossless 1 input.png output_lossless.jxl

# Lossless JPEG transcoding (mathematically lossless)
cjxl --lossless_jpeg=1 input.jpg output_from_jpeg.jxl

# Progressive (for web streaming)
cjxl --progressive 1 --quality 80 input.jpg output_progressive.jxl

# Decoding JXL
djxl output.jxl decoded.png  # or .jpg, .ppm, etc.

ffmpeg

# JXL via ffmpeg (requires libjxl support compiled in)
ffmpeg -i input.jpg -c:v libjxl -q:v 80 output.jxl

ImageMagick (7.1+)

convert input.jpg -quality 80 output.jxl

Browser Support Situation (2024)

Browser	Status
Safari / iOS Safari	✅ Native support (17+)
Chrome	❌ Removed Aug 2022, flag removed 2023; working on re-enabling
Firefox	❌ Behind flag (media.jxl.enabled=true in about:config)
Edge	❌ Not supported (Chromium-based, follows Chrome)

Chrome's removal of JPEG XL in August 2022 was highly controversial in the web standards community and generated significant backlash. The stated reason was insufficient ecosystem adoption and resource constraints. The format remains actively developed and specification-complete.

Practical implication: JPEG XL cannot currently be served directly to Chrome users without fallback. Use it for:

Safari-specific delivery (iOS App Store apps, Apple ecosystem)
Server-side storage (transcode to WebP/AVIF for delivery to non-Safari browsers)
Image archives and lossless JPEG migration pipelines

JPEG XL vs. AVIF — When to Use Each

Criterion	Choose JXL	Choose AVIF
Chrome support needed	❌	✅
Maximum compression	✅ (slightly better)	✅ (close)
Decoding speed	✅ (faster)	Medium
Lossless JPEG migration	✅ (unique feature)	❌
32-bit/HDR/CMYK	✅	Limited
Progressive decoding	✅	❌
Animation (Safari)	✅	❌ (Safari doesn't support animated AVIF)
Broad browser support	❌	✅

Summary

JPEG XL is technically the most capable image format available — it outperforms AVIF in compression at high quality, adds lossless JPEG transcoding, achieves faster decoding, and supports HDR up to 32-bit float. Its limitation is browser adoption: Chrome's controversial decision to remove support has stalled web deployment. For professional image workflows, lossless JPEG archives, and Safari-targeted content, JXL is compelling. For broad web deployment today, AVIF with WebP fallback remains the practical choice. Monitor Chrome's JXL re-enabling status — if Chrome adds support, JXL could become the universal modern image format.

Related conversions

Most teams that read this guide convert images in one of these directions:

Related conversions

Put what you just learned into practice — convert your files now in seconds, free and without registration.

Convert →

Convert →

Convert →

Convert →

Convert →

Convert →

Frequently Asked Questions

Google removed JPEG XL support from Chrome in August 2022 (Chrome 110), citing insufficient ecosystem adoption beyond Chrome itself and concerns about the technical cost of maintaining a non-standard codec. This decision was highly controversial — JPEG XL had wide support from image professionals, CDN providers (Cloudflare, Fastly), and the format's technical superiority was not disputed. Safari added native JXL support in 2023. Chrome has indicated it may revisit the decision, and the JPEG XL specification remains ISO-standardized (ISO/IEC 18181) and actively maintained.

Use <strong>lossy</strong> (JPG, WebP, AVIF) for photographs — the human eye barely notices the difference at quality 80–85, and file sizes are 5–20× smaller. Use <strong>lossless</strong> (PNG, WebP-lossless) for screenshots, UI mockups, logos and anything with sharp edges or text — lossy creates ugly artifacts around boundaries.

JPEG XL can store the exact DCT coefficients from a JPEG file along with JPEG metadata, creating a JXL file that: (1) is mathematically lossless — converting back to JPEG produces a bit-for-bit identical file, (2) is typically 20% smaller than the original JPEG, and (3) requires no quality decisions. Command: cjxl --lossless_jpeg=1 input.jpg output.jxl. This enables migrating entire JPEG archives (photos, web assets) to JXL without any quality loss — a unique feature no other format offers.

JPEG XL can store the exact DCT coefficients de a JPEG arquivo along com JPEG metadata, creating a JXL arquivo that: (1) is mathematically sem perdas — convertendo back to JPEG produces a bit-for-bit identical file, (2) is tipicamente 20% menor que the original JPEG, e (3) requires no quality decisions. Command: cjxl --lossless_jpeg=1 input.jpg output.jxl. This enables migrating entire JPEG archives (photos, web assets) to JXL sem any quality loss — a unique feature no other formato offers.

JPEG XL can store the exact DCT coefficients von a JPEG Datei along mit JPEG metadata, creating a JXL Datei that: (1) is mathematically verlustfrei — umwandelnd back to JPEG produces a bit-for-bit identical file, (2) is typically 20% kleiner als the original JPEG, und (3) requires no quality decisions. Command: cjxl --lossless_jpeg=1 input.jpg output.jxl. This enables migrating entire JPEG archives (photos, web assets) to JXL ohne any quality loss — a unique feature no other Format offers.

JPEG XL can store the exact DCT coefficients de a JPEG archivo along con JPEG metadata, creating a JXL archivo that: (1) is mathematically sin pérdidas — convirtiendo back to JPEG produces a bit-for-bit identical file, (2) is typically 20% más pequeño que the original JPEG, y (3) requires no quality decisions. Command: cjxl --lossless_jpeg=1 input.jpg output.jxl. This enables migrating entire JPEG archives (photos, web assets) to JXL sin any quality loss — a unique feature no other formato offers.

For new projects in 2026, WebP is supported by all modern browsers (95%+ of traffic) and saves 25–35% over JPG at the same visual quality. AVIF is even more efficient (40–50% smaller) but encoding is slower and Safari support is recent. Use WebP as the default and AVIF as the progressive enhancement via <code><picture></code> with JPG fallback.

On technical merits, JPEG XL edges out AVIF: it achieves slightly better compression at high quality settings, decodes 2-4x faster, supports lossless JPEG transcoding (unique), handles 32-bit float and CMYK (impossible in AVIF), and supports progressive decoding. However, AVIF has 95%+ browser support including Chrome while JPEG XL lacks Chrome support. For web delivery requiring Chrome compatibility, AVIF wins on practical grounds. For professional workflows, archival, and Safari-targeted content, JPEG XL is technically superior.

On technical merits, JPEG XL edges out AVIF: it achieves slightly better compressão at alta qualidade settings, decodes 2-4x faster, suporta sem perdas JPEG transcoding (unique), handles 32-bit float e CMYK (impossible in AVIF), e suporta progressive decoding. Porém, AVIF has 95%+ browser support including Chrome while JPEG XL lacks Chrome support. para web delivery requiring Chrome compatibilidade, AVIF wins on practical grounds. para profissional workflows, archival, e Safari-targeted content, JPEG XL is technically superior.

On technical merits, JPEG XL edges out AVIF: it achieves slightly better Komprimierung at hohe Qualität settings, decodes 2-4x faster, unterstützt verlustfrei JPEG transcoding (unique), handles 32-bit float und CMYK (impossible in AVIF), und unterstützt progressive decoding. Jedoch, AVIF has 95%+ browser support including Chrome while JPEG XL lacks Chrome support. für web delivery requiring Chrome Kompatibilität, AVIF wins on practical grounds. für professionell workflows, archival, und Safari-targeted content, JPEG XL is technically superior.

On technical merits, JPEG XL edges out AVIF: it achieves slightly better compresión at alta calidad settings, decodes 2-4x faster, soporta sin pérdidas JPEG transcoding (unique), handles 32-bit float y CMYK (impossible in AVIF), y soporta progressive decoding. Sin embargo, AVIF has 95%+ browser support including Chrome while JPEG XL lacks Chrome support. para web delivery requiring Chrome compatibilidad, AVIF wins on practical grounds. para profesional workflows, archival, y Safari-targeted content, JPEG XL is technically superior.

It depends on the tool: ImageMagick and FFmpeg copy EXIF by default, while cwebp drops it unless you pass <code>-metadata all</code>. KaijuConverter strips metadata server-side after conversion to protect your privacy — if you need to keep camera or GPS data, use a desktop tool you control.

Conditionally. If your audience is primarily Safari/iOS users, JPEG XL is viable with WebP fallback using the <picture> element. If you need Chrome support, serve AVIF or WebP to Chrome users and JXL to Safari users via server-side content negotiation (checking the Accept: image/jxl header). For image archives and lossless JPEG migration pipelines running server-side, JPEG XL is excellent regardless of browser support. Use tools like cjxl/djxl for conversion in batch processing workflows.

Conditionally. If your audience is primarily Safari/iOS users, JPEG XL is viable com WebP fallback usando the <picture> element. If you need Chrome support, serve AVIF ou WebP to Chrome users e JXL to Safari users via server-side content negotiation (checking the Accept: image/jxl header). para image archives e sem perdas JPEG migration pipelines running server-side, JPEG XL is excellent regardless of browser support. usar ferramentas like cjxl/djxl para conversion in batch processing workflows.

Conditionally. If your audience is primarily Safari/iOS users, JPEG XL is viable mit WebP fallback using the <picture> element. If you need Chrome support, serve AVIF oder WebP to Chrome users und JXL to Safari users via server-side content negotiation (checking the Accept: image/jxl header). für image archives und verlustfrei JPEG migration pipelines running server-side, JPEG XL is excellent regardless von browser support. verwenden Werkzeuge like cjxl/djxl für conversion in batch processing workflows.

Conditionally. If your audience is primarily Safari/iOS users, JPEG XL is viable con WebP fallback using the <picture> element. If you need Chrome support, serve AVIF o WebP to Chrome users y JXL to Safari users via server-side content negotiation (checking the Accept: image/jxl header). para image archives y sin pérdidas JPEG migration pipelines running server-side, JPEG XL is excellent regardless de browser support. usar herramientas like cjxl/djxl para conversion in batch processing workflows.

Three common causes: (1) ICC color profiles dropped during conversion shift colors slightly; (2) chroma subsampling (4:2:0) reduces color accuracy in lossy JPG/WebP; (3) the new format may not support all features of the original (e.g. PNG → JPG drops transparency to white). For pixel-perfect results stick to lossless formats and preserve color profiles explicitly.