photomopOpen the studio

Compression

Chroma subsampling

Also called 4:2:0, 4:4:4 or chroma sub-sampling. Here is what it means, when it changes what you export, and what people get wrong.

CompressionRuns on your deviceUpdated 10 August 2026

The short answer

Chroma subsampling stores colour information at a lower resolution than brightness, because human vision resolves fine detail in brightness far better than in colour. Written as ratios such as 4:4:4, 4:2:2 and 4:2:0, it cuts file size cheaply on photographs but smears saturated edges, coloured text and thin graphics.

Updated

The short version

Chroma subsampling at a glance

Quick facts about Chroma subsampling: notation, full colour, web default, colour data dropped at 4:2:0, applies to
Quick factDetail
NotationRegion width : top row samples : bottom row samples
Full colour4:4:4, one chroma sample per pixel
Web default4:2:0 in most JPEG, WebP and AVIF encoders
Colour data dropped at 4:2:0Three quarters of the chroma samples
Applies toLossy codecs, not PNG or lossless modes

Slide the table sideways to see every column.

How does it actually save space?

Before compressing, a JPEG or WebP encoder converts the image out of RGB into a luma plus chroma representation: one channel carrying brightness, two carrying colour difference. Those three channels can then be treated separately.

Your eye has far more receptors for light level than for colour, and the visual system pools colour information over larger areas. So the encoder keeps the brightness channel at full resolution and shrinks the two colour channels, often to half width and half height. That discards three quarters of the colour samples before any quantisation happens at all, and on most photographs you cannot see it.

How do you read 4:2:0 and 4:4:4?

The notation describes a reference region four pixels wide and two pixels tall. The first number is the width of that region, the second is how many colour samples are taken across the top row, and the third is how many additional colour samples are taken across the bottom row. Zero in the third position means the bottom row reuses the top row's colour.

ModeColour resolutionTypical use
4:4:4Full, one colour sample per pixelGraphics, text, logos, screenshots, masters
4:2:2Half horizontally, full verticallyVideo capture and editing, a middle ground
4:2:0Half horizontally and verticallyDefault for web JPEG, WebP, AVIF, most cameras

Most encoders switch to 4:4:4 automatically at their highest quality settings, but the threshold varies by tool, so check rather than assume.

Why does 4:2:0 wreck red text on white?

Because that edge lives almost entirely in the colour channels. Bright red and white are not far apart in brightness, so the luma channel, the one kept at full resolution, records only a faint change. Nearly all the information defining the edge sits in the chroma channels, which have just been halved in both directions. The result is a soft, smeared, sometimes fringed edge that no amount of raising the quality slider will sharpen.

The same failure hits thin coloured lines, saturated logos on pale backgrounds, price badges, packaging small print and any screenshot of coloured UI. Force 4:4:4 for those, or step outside lossy coding entirely and use PNG or lossless WebP. For ordinary product photography and general photographs, 4:2:0 is close to free and worth keeping.

What people get wrong about chroma subsampling

Each one is a real failure mode, not a style preference.

  1. Raising the JPEG quality to 95 to fix blurry red text, when the damage comes from 4:2:0 chroma and needs 4:4:4 instead.

  2. Assuming 4:4:4 costs a lot of file size on ordinary photographs, when the difference is often modest and worth it for graphics.

  3. Exporting screenshots and UI mockups through a default web encoder, which subsamples coloured interface elements into mush.

  4. Believing the ratio numbers describe colour depth or bits per channel rather than sampling positions.

Questions people ask

Chroma subsampling, answered

The follow-up questions people search for once they have the definition.

Should I always use 4:4:4?

No. For photographs it usually adds size for no visible gain, since your eye cannot resolve the extra colour detail. Use 4:4:4 when the image contains coloured text, logos, thin graphic lines or screenshots, where the edges depend on chroma resolution.

Does PNG use chroma subsampling?

No. PNG stores full RGB values for every pixel and is lossless, so there is no chroma plane to subsample. The same is true of lossless WebP. Subsampling belongs to lossy codecs such as JPEG, WebP lossy, AVIF and video formats.

Can you tell which subsampling a JPEG used?

Yes. Most image inspection tools and command line utilities report the sampling factors from the file's headers. Visually, look at a saturated coloured edge against a light background at full zoom, where 4:2:0 shows a soft blocky fringe.

Why do my product photos look fine but the packaging text does not?

The photograph is full of texture that hides the loss, while printed text is a hard high-contrast edge, often in a brand colour. Subsampling halves the colour resolution defining that edge, so the small print goes soft while everything around it looks acceptable.

From the studio

Tools that do this

Each one runs on your own device, so the photo stays where it is.

Sources

Where the facts on this page come from. Every link opens the specification, standard or documentation the claim was read at.

Now do it to a photo

Photomop is a photo studio that runs on your own device. Resize, crop, compress, convert, batch edit, remove a background and change one all work in the browser tab you are reading this in, at full resolution and with no watermark. The on-device page shows you how to check that the photo stays put.