How to Compress JPG: Quality Settings Explained

JPEG compression is controlled primarily by a quality parameter that trades file size against image fidelity. jpg.now exposes this slider with named presets to make the choice intuitive, but understanding what's happening under the hood helps you pick the right setting for each use case.

The Quality Slider (0–100)

The quality value controls the quantisation table used during DCT (Discrete Cosine Transform) compression. Higher values preserve more frequency coefficients, keeping fine detail at the cost of a larger file. The relationship between quality and file size is non-linear: dropping from 95 to 85 often reduces file size by 40–50% with barely perceptible quality loss, while dropping from 85 to 60 produces diminishing returns on quality.

Recommended Presets

PresetQualityBest for
Small60Thumbnails, email attachments, where size dominates
Balanced82Web images, social media - Good size/quality ratio
High90Portfolio sites, client deliverables
Max96Near-lossless; archiving when storage is available
Lossless-Uses jpegtran lossless optimisation; no quality change

Chroma Subsampling

Human vision is less sensitive to colour (chrominance) detail than to brightness (luminance). JPEG exploits this by storing colour at lower resolution than brightness- A technique called chroma subsampling. The most common scheme, 4:2:0, halves the colour resolution horizontally and vertically, reducing file size by ~15% with no visible impact on photographs. For text-heavy images or graphics, 4:4:4 (no subsampling) avoids colour fringing on sharp edges.

Progressive JPEG

A progressive JPEG encodes the image in multiple passes from low to high frequency. On slow connections, the browser shows a blurry preview that sharpens as more data arrives, rather than a top-to-bottom scan. Progressive encoding also produces files 2–5% smaller than baseline JPEG at the same quality, making it worth enabling for web use.

Lossless Mode

The Lossless preset uses jpegtran-style optimisation: it reorders and optimises the Huffman coding tables in your existing JPEG without decoding and re-encoding pixel data. Typical savings are 5–15%. No visual quality change occurs because the pixel values are never altered. This is the right choice when you have a high-quality master JPEG and just want to reduce its size without any generation loss.

Ready to convert?
Free, no signup required. Files deleted in 24 h.
Start converting
Back to all FAQ