How JPEG Compression Actually Works
JPEG compression works by discarding detail that the human eye is unlikely to notice. It splits the image into 8×8 pixel blocks and applies a mathematical transform (Discrete Cosine Transform) to each block. The compression level decides how much high-frequency detail — fine textures, sharp edges, subtle color gradients — gets discarded.
At quality 100%, you keep nearly everything. At quality 50%, you have thrown out half the detail. But here is the trick: your eye cannot resolve the difference between 100% and 85% on a typical photograph viewed at normal size. The file size difference, however, is massive — often 5-10×.
| Quality | Typical Size (vs Original) | Visible Difference? | Use Case |
|---|---|---|---|
| 100% | 100% (baseline) | None | Archive copy |
| 92% | ~40-50% | None for most images | Print, retina screens |
| 85% | ~25-35% | None at normal viewing | Web images (recommended) |
| 75% | ~15-25% | Minor — zoom to see | Heavy web pages, email |
| 60% | ~10-18% | Noticeable on close inspection | Thumbnails, previews |
| 40% | ~5-10% | Obvious artifacts | Only for extreme size constraints |
These are averages from my own photo library — 50 JPEGs at 1920×1280. I was genuinely surprised that quality 85 and 100 were indistinguishable at normal viewing size. The one exception: images with fine black text on white backgrounds. Those show ringing artifacts around the text at anything below 92. For those, use PNG or WebP instead — JPEG is the wrong tool for text.
Chroma Subsampling: The Hidden Quality Killer
Chroma subsampling is a separate JPEG setting that reduces color resolution while keeping brightness at full resolution. Since the human eye is more sensitive to brightness than color, this is free file size reduction — up to 25% smaller with no visible difference.
Most JPEG encoders default to 4:2:0 subsampling (half color resolution both ways), which is fine for photos. But avoid 4:2:0 for images containing sharp colored text or thin red/blue lines — it can cause visible color bleeding. In that case, use 4:4:4 (no subsampling) and reduce the quality setting instead.
PicFix's compressor uses the browser's built-in encoder, which defaults to 4:2:0 subsampling for JPEG — correct for 95% of photos. For fine text or graphics, use PNG or WebP instead.
Progressive JPEG: Better Perceived Speed
Progressive JPEG loads in passes — first a blurry full-size preview, then progressively sharper. A baseline JPEG loads top-to-bottom. For images over 100KB, progressive JPEG improves perceived page speed because users see the content immediately instead of watching it render line by line.
Progressive JPEGs are typically 2-5% smaller than baseline JPEGs at equivalent quality. There is no quality trade-off. The only downside: older decoders (pre-2010) may not support them. In 2026, this is irrelevant — every browser handles progressive JPEG perfectly.
Strip Metadata for Extra Savings
JPEGs carry metadata — EXIF (camera settings, GPS location, date), XMP (editing history), IPTC (copyright), and ICC color profiles. This data can add 10-50KB to a file. For web images, you should strip everything except the ICC profile (which ensures correct color display).
Use PicFix's EXIF viewer and stripper to see exactly what metadata your image carries and remove it before publishing. This is also a privacy best practice — EXIF data can contain GPS coordinates of where the photo was taken.
The 3-Step Compression Workflow
- Resize to display dimensions. If your image will appear at 800px wide on your website, resize it to exactly 800px (or 1600px for high-DPI screens). Do not serve a 4000px image to a 400px slot. Use the resize tool.
- Compress at quality 85%. For JPEG, this is the sweet spot. Visually indistinguishable from 100% for photographs at normal viewing distance. Use the compress tool with the quality slider.
- Compare side by side. Always visually confirm. If your image has fine text or sharp color transitions, you may need quality 92-95%. If it is a smooth gradient or soft-focus photo, you can go as low as 70%. Trust your eyes, not a number.
When NOT to Use JPEG
JPEG is the wrong format for:
- Screenshots with text. JPEG compression creates ringing artifacts around sharp edges. Use PNG or WebP instead.
- Logos and icons. These are typically flat colors with sharp edges — JPEG smoothes them. Use SVG first, PNG second.
- Images needing transparency. JPEG has no alpha channel. Use PNG (lossless) or WebP (lossy, but supports transparency).
- Images you will edit again. Every JPEG save introduces new compression artifacts. Edit in a lossless format (PNG, TIFF), export to JPEG only as the final step.
Compress your JPEGs now
Drop an image into PicFix's compressor and watch the file size drop in real time as you move the quality slider. All processing in your browser — your files never touch a server.