Spreadsheet with data, the practical side of image compression numbers
Optimization

How to Compress Images Without Visible Quality Loss

The scenario is always the same: you hit "upload" on a form, a CMS, an email attachment or a marketplace listing, and you're told the file is too big. Your 12-megapixel photo straight out of a camera or phone is 4–6 MB. The portal wants 2 MB. The email wants 5 MB total. And the last time you used a "compress" tool, the result looked like a watercolor painting. This guide is about getting from 5 MB to 300–500 KB with nobody able to tell the difference — and understanding which dial actually matters.

Why you're being forced to compress in the first place

A photo file size is the product of three things: pixels (resolution), bit depth, and compression quality. Modern phones throw away almost nothing on the third knob — they save at near-maximum JPEG quality, around 95–98, because storage was the bottleneck for the last decade, not file size. Now storage is cheap, but web forms, mail servers and social platforms are still built for 2005. So your 5 MB masterpiece gets refused or downscaled by someone else's server, often badly.

The fix is not "make the quality lower until it fits" blindly. It is to separate the two knobs — resolution and quality — and use the right one. Most people reach for quality first, when the real waste is usually resolution.

Why quality 80% is the sweet spot

JPEG quality is not linear — it is a perceptual scale disguised as a number. The difference between quality 100 and 90 is enormous in file size and invisible to the eye. The difference between 80 and 70 starts to show on close inspection. Between 60 and 40, anyone can see it immediately.

The reason is how JPEG compression spends its budget. At 90+, it wastes bits preserving near-imperceptible noise — sensor grain, the texture of a clear blue sky, dithering artifacts that came from compression in the first place. At 80, it drops exactly those bits while keeping everything the eye actually uses: edges, skin tones, color, and detail. This is why the entire web — Facebook, Instagram, WordPress, Shopify — re-compresses your uploads to roughly this range regardless of what you gave them. They know the math.

Practical rule: export web and upload images at quality 80–85. It is the biggest free win in all of image optimization, and the "loss" is something you cannot see on a screen.

What chroma subsampling is doing behind the slider

Here's the detail most compression tools hide: JPEG doesn't compress brightness and color equally. Human vision is far more sensitive to luma (light/dark) than to chroma (hue/saturation). So JPEG throws away color resolution first — typically storing color at half or quarter the resolution of brightness. This is called chroma subsampling, written as ratios like 4:4:4 (no reduction), 4:2:2 (half) and 4:2:0 (quarter).

  • 4:2:0 — default for cameras and the web. Half the horizontal, half the vertical color detail. Look at a bright red logo on white, zoom in: edges look soft and slightly muddy. Fine for photos, bad for text, logos, and flat color blocks.
  • 4:2:2 / 4:4:4 — needed when your image has sharp text overlays, line art, or screenshots. Crank the quality or use PNG for those instead — a screenshot saved as JPEG at 4:2:0 will show ringing around letters no matter how high the quality slider is.

This is why screenshots compress like garbage as JPEG. If your image has text or hard edges, the right move is PNG (lossless, handles edges), not a higher JPEG quality number.

Advertisement

Real measurement: same photo, every quality level

To remove the guesswork, I took a typical product photo (20 MP camera, outdoor, complex foliage background) and re-encoded it at every quality step, keeping the resolution fixed at 2000 px wide:

JPEG qualityFile size (2000 px wide)Perceptual resultWhere to use it
1003.9 MBMathematically lossless-looking; wastefulNever needed on the web
901.4 MBIndistinguishable from original on screenPrint masters, high-end portfolio
80720 KBNo visible loss on normal viewingDefault: web, listings, forms, email
60380 KBSoftness on close crops; visible banding in skiesThumbnails, low-bandwidth previews
40210 KBBlocky artifacts, muddy colors, "compressed" lookAvoid

Going from 100 to 80 saved 81% of the bytes for zero perceptible loss. That is the single highest-return optimization in this entire subject. Going from 80 to 60 saved only another 47% of the already-small file, and introduced visible quality problems. Compress until you hit the target, but stop at 80 unless you're desperate.

If the target is still not met at quality 80, the answer is rarely quality 60. It is the other knob.

When to resize instead of recompress

A 20 MP photo displayed in a form at 800 px wide is being shrunk by the recipient's software anyway. Why carry 5472 × 3648 pixels to show 800 × 533? Resolution is the bigger lever, and unlike quality, reducing resolution only costs detail nobody sees.

  • Web uploads, listings, social — 1600 px on the long edge is plenty. A 1600 px-wide JPEG at quality 80 is usually 250–450 KB.
  • Email attachments — 1200 px on the long edge, quality 80. This gets a photo under 300 KB.
  • Document scans / PDFs for upload — 150 DPI is fine for screen viewing; reserve 300 DPI for printing.

Resizing a 5 MB image down to 1600 px wide and saving at quality 80 typically lands at 300–500 KB — roughly a 90% reduction — with better apparent quality than keeping the full resolution at quality 50. Pick resolution first, quality second.

One more trap: never keep re-saving the same JPEG. Every save re-compresses, and quality compounds downward. Always edit from the original master, or convert once from the original. Our WebP converter and the format converter both re-encode from the source file in one pass, so you make exactly one lossy generation instead of five.

Need a smaller photo that still looks sharp? Convert and re-encode images in your browser at quality 80 — the sweet spot, applied automatically.

Open the image format converter →

Frequently asked questions

Is there such a thing as "lossless" JPEG compression?

JPEG is inherently lossy. "Lossless compression" tools that shrink a JPEG by 5–15% only strip metadata and optimize the Huffman tables — the visible pixels are already lossy. Bigger wins need re-encoding at quality 80–90 or reducing resolution.

Why does my screenshot look terrible when saved as JPEG?

JPEG's 4:2:0 chroma subsampling and block-based compression smear sharp edges and text. Screenshots, logos and graphics with text should be PNG (lossless) or WebP, not JPEG.

What file size counts as "compressed enough"?

For most web forms and listings, under 1–2 MB. For email attachments, aim under 500 KB. For thumbnails and previews, under 150 KB. Start from a 1600 px-wide export at quality 80 and adjust from there.

Does PNG compress better than JPEG for photos?

No — the opposite. PNG is lossless, so a photo as PNG is usually 3–8× larger than the same photo as JPEG at quality 80. PNG is only correct for screenshots, transparency and graphics; JPEG (or WebP) for photos.

Will compressing reduce the quality enough to matter for printing?

Quality 80 at 300 DPI prints fine up to about A4/Letter. For larger posters, keep a quality 90–95 master. Compression aimed at the web (1600 px) will look soft if stretched to poster size — keep the original full-resolution file as the print master.

Advertisement

Related reading