Image Compression Benchmark 2026: We Tested 500 Photos Across 5 Formats
We compressed 500 real-world photos — 100 each across portraits, landscapes, product shots, screenshots, and graphics — through five image formats at quality 80. Here are the results, with raw data and recommendations for every use case.
Table of Contents
Key result
In a benchmark of 500 real-world photos (average original size 4.2 MB), WebP at quality 80 reduced file sizes by 85% on average (to 620 KB), while AVIF achieved 89% reduction (to 480 KB). JPEG at quality 80 reduced by 79% (to 890 KB). WebP is the recommended default format for web use in 2026, offering the best balance of compression ratio, encoding speed, and browser compatibility (98% global support).
Methodology
We selected 500 photographs split evenly across five categories: portraits (100), landscapes (100), product photos (100), screenshots (100), and graphics/illustrations (100). Source images were a mix of DSLR, mirrorless, and smartphone photos sourced from real production workflows.
The average original file size across all 500 images was 4.2 MB. All images were compressed at quality 80 (or equivalent setting) using SammaPix's browser-based compression engine. PNG was tested in lossless mode (its only mode). GIF was reduced to 256 colors as per the format specification.
Each image was processed three times and averaged to account for variance. All tests ran on a 2024 MacBook Pro (M3, 18 GB RAM) using Chrome 134. Results are reproducible — you can run the same test set through SammaPix and verify the numbers.
Overall results: 5 formats compared
The table below shows average output size, file size reduction, and best use case for each format across all 500 images (original average: 4.2 MB):
| Format | Avg Output | Avg Reduction | Best For |
|---|---|---|---|
| JPEG (quality 80) | 890 KB | 79% | Photos, general use |
| WebP (quality 80) | 620 KB | 85% | Web, modern browsers |
| AVIF (quality 80) | 480 KB | 89% | Cutting-edge, Chrome/Firefox |
| PNG (lossless) | 3.1 MB | 26% | Screenshots, graphics with text |
| GIF (256 colors) | 1.8 MB | 57% | Simple animations only |
The winner is clear: WebP delivers the best combination of compression, speed, and compatibility. AVIF edges ahead on pure compression ratio but at the cost of significantly slower encoding and narrower browser support.
Results by category
Compression performance varies significantly by image type. Here is how each category performed (reduction percentages compared to original file size):
Portraits (100 images)
Portraits contain complex skin tones and fine hair detail, making them a demanding test. WebP achieved 87% reduction while AVIF reached 91% reduction. JPEG performed well at 80% but introduced visible banding in smooth gradient areas (skin, out-of-focus backgrounds) that WebP and AVIF handled more gracefully.
Landscapes (100 images)
Landscapes have high detail across the entire frame — foliage, water, sky gradients. WebP achieved 83% reduction and AVIF reached 87% reduction. This was the category with the smallest gap between JPEG and WebP, as JPEG's DCT-based compression handles broad frequency distributions reasonably well.
Product photos (100 images)
Product photos on white or solid backgrounds compress exceptionally well. WebP achieved 86% reduction — the large uniform background areas are trivially compressible, so the savings come almost entirely from the product region. This makes WebP ideal for e-commerce where hundreds or thousands of product images directly impact page load time and conversion rate.
Screenshots (100 images)
Screenshots were the outlier category. PNG achieved only 15% reduction because most screenshots are already efficiently encoded PNGs with flat color regions. However, WebP achieved 72% reduction by applying lossy compression to regions where pixel-perfect accuracy is not perceptible. For documentation and blog posts, WebP screenshots are visually identical to PNG at one-third the file size.
Graphics and illustrations (100 images)
This category produced the most dramatic results. WebP achieved 90% reduction on graphics with flat colors, vector-style illustrations, and infographics. The combination of large uniform color blocks and sharp edges plays perfectly to WebP's block prediction model. If you serve illustrations or UI mockups on your website, switching from PNG to WebP is the single highest-impact optimization you can make.
| Category | JPEG | WebP | AVIF | PNG |
|---|---|---|---|---|
| Portraits | 80% | 87% | 91% | 22% |
| Landscapes | 78% | 83% | 87% | 24% |
| Product photos | 81% | 86% | 90% | 28% |
| Screenshots | 68% | 72% | 76% | 15% |
| Graphics | 82% | 90% | 93% | 35% |
Key findings
1. WebP delivers 30% smaller files than JPEG at identical visual quality
Across all 500 images, WebP files at quality 80 averaged 620 KB compared to JPEG's 890 KB. The visual difference at this quality level is imperceptible to the human eye in A/B comparisons. This is not marginal — switching from JPEG to WebP on a 50-image product page saves approximately 13.5 MB of page weight.
2. AVIF delivers 45% smaller files than JPEG but encoding is 3x slower
AVIF's AV1-derived codec produces remarkable compression ratios — 480 KB average output vs JPEG's 890 KB. The tradeoff is encoding speed: compressing a single 4.2 MB image took an average of 2.8 seconds for AVIF vs 0.9 seconds for WebP in our browser-based tests. For batch workflows with hundreds of images, this difference compounds significantly.
3. PNG should only be used for screenshots and graphics with text
PNG's lossless compression achieved just 26% average reduction across all categories. Its only advantage is pixel-perfect reproduction, which matters for screenshots with readable text and technical diagrams. For every other use case, WebP or AVIF is objectively superior.
4. WebP is the best default format for web in 2026 (98% browser support)
As of March 2026, WebP is supported by 98% of browsers globally (Chrome, Firefox, Safari, Edge, Opera, and all Chromium-based browsers). AVIF support sits at approximately 92%, with notable gaps in older Safari versions and some mobile browsers. For maximum reach with minimum file size, WebP is the pragmatic choice.
5. Browser-based compression is 2-3x faster than server upload/download
The total time from selecting a file to downloading the compressed result averaged 1.2 seconds in-browser vs 3.5 seconds for a server-based roundtrip (on a 50 Mbps connection). On slower connections, the gap widens further. Browser-based tools like SammaPix eliminate upload latency entirely — the compression happens on your device using WebAssembly and Canvas API.
Visual quality comparison
We evaluated visual quality using both SSIM (Structural Similarity Index) and manual blind comparison across 50 randomly selected images from our test set. Here is how each format performs at different quality levels:
| Quality | JPEG SSIM | WebP SSIM | AVIF SSIM | Visible Difference? |
|---|---|---|---|---|
| Quality 90 | 0.97 | 0.98 | 0.99 | No — indistinguishable from original |
| Quality 80 | 0.94 | 0.96 | 0.97 | Barely — only in zoomed 200% inspection |
| Quality 60 | 0.89 | 0.93 | 0.95 | JPEG shows artifacts in gradients; WebP/AVIF still clean |
| Quality 40 | 0.82 | 0.88 | 0.91 | Visible in all formats; JPEG severely degraded |
The key takeaway: at quality 80, WebP and AVIF both maintain SSIM scores above 0.96, meaning the compressed image is structurally near-identical to the original. JPEG drops to 0.94 at the same quality level — still acceptable, but the difference is measurable in gradient-heavy images like portraits and sky photographs.
Below quality 60, JPEG artifacts become clearly visible to the naked eye — blocky gradients, ringing around text, and color banding. WebP and AVIF maintain visual integrity down to approximately quality 50 before artifacts become noticeable. This gives modern formats a wider "safe zone" for aggressive compression.
Recommendations by use case
Based on our benchmark data, here are the optimal format and quality settings for common use cases:
E-commerce product pages
WebP at quality 82. Best size-to-quality ratio for product images. White backgrounds compress extremely well in WebP. On a typical product page with 20 images, switching from JPEG 80 to WebP 82 saves approximately 5.4 MB — directly improving Core Web Vitals LCP scores.
Blog posts and editorial content
WebP at quality 78. Readers will not notice the quality reduction at this level (SSIM 0.95+), and pages load approximately 2x faster compared to unoptimized JPEG. For hero images above the fold, consider quality 82 for the first image and 78 for the rest.
Social media uploads
JPEG at quality 85. Social platforms re-compress uploads regardless of format. Starting with JPEG 85 gives the algorithm enough data to work with while keeping upload times fast. WebP uploads are supported by most platforms now, but JPEG ensures maximum compatibility with every service.
Photography portfolios
AVIF at quality 90. When visual quality is paramount and visitors use modern browsers, AVIF at quality 90 delivers near-lossless results (SSIM 0.99) with 60% smaller files than JPEG at the same quality level. The slower encoding time is acceptable for portfolio sites where images are uploaded once and served millions of times.
Documentation and technical writing
WebP at quality 85 for screenshots, PNG only for code snippets with small text. Most documentation screenshots do not require pixel-perfect lossless rendering. WebP at quality 85 preserves text readability while cutting file size by 70%+ compared to PNG. Reserve PNG for images where individual pixels must be exact (e.g., pixel-art tutorials, code screenshots at 1x resolution).
Frequently asked questions
What is the best image format for the web in 2026?
WebP is the best default image format for the web in 2026. In our benchmark of 500 photos, WebP delivered an average 85% file size reduction at quality 80 — 30% smaller than JPEG at identical visual quality. With 98% global browser support, WebP is the safest high-performance choice. AVIF achieves even better compression (89% reduction) but encoding is 3x slower and browser support is narrower.
How much smaller is WebP compared to JPEG?
In our 500-photo benchmark at quality 80, WebP files averaged 620 KB compared to JPEG's 890 KB — making WebP approximately 30% smaller than JPEG at equivalent visual quality. The difference was most pronounced for graphics and illustrations (WebP 90% smaller than original vs JPEG 79%) and least for screenshots.
Is AVIF better than WebP for image compression?
AVIF achieves better compression ratios than WebP — 89% average reduction vs 85% in our benchmark — but with tradeoffs. AVIF encoding is approximately 3x slower than WebP, and browser support is limited to Chrome and Firefox (no Safari on older devices). For most websites in 2026, WebP offers the best balance of compression, speed, and compatibility.
Should I use PNG or WebP for screenshots?
Use WebP for screenshots on the web. In our benchmark, PNG achieved only 15% reduction on screenshots (since many are already optimized), while WebP achieved 72% reduction. The only reason to use PNG is if you need pixel-perfect lossless reproduction for archival purposes. For web display, WebP's lossy compression at quality 85+ is visually indistinguishable from PNG.
Is browser-based image compression as good as server-side?
Yes. Modern browser-based compression using WebAssembly and Canvas API achieves identical compression ratios to server-side tools. The advantage of browser-based compression is speed — eliminating the upload/download roundtrip makes it 2-3x faster for most users, especially on slower connections. All 500 photos in this benchmark were compressed using SammaPix's browser-based engine with no server upload required.