JPEG Quality vs File Size: What the Slider Really Does

Open any image editor, export a photo at quality 90, then export the same photo at quality 80. The second file is often a third smaller, and you cannot see the difference. Export it again at 60 and the file barely shrinks further, but the photo starts to look rough. That uneven trade is the whole story of JPEG quality vs file size, and once you see the shape of it, the slider stops feeling random.

What the JPEG quality number controls

The quality number controls how aggressively the encoder throws away detail. A JPEG stores an image as blocks of frequency data, and the quality setting picks a quantization table that decides how much of that data to keep. A high number keeps more detail and makes a bigger file. A low number discards more and makes a smaller one.

The number runs from 1 to 100 in most editors, but the scale is not a percentage of anything. Quality 50 does not mean half the detail is gone. It means the encoder picked the middle table from a set that the JPEG standard defined decades ago, and the tables at the top end are much closer together than the ones at the bottom.

JPEG quality vs file size: why the curve bends

JPEG quality vs file size follows a curve that flattens fast. Dropping from 95 to 85 can cut a file by a third or more, because the encoder is discarding detail you barely notice. Dropping from 60 to 50 saves far less, because most of the discardable detail is already gone, and what remains is the structure your eyes see.

Rough figures for a typical 12-megapixel phone photo. Yours will differ, because content matters more than the setting, but the shape of the curve holds for nearly every photo.

Quality	Rough file size
95	5-7 MB
90	3-4 MB
85	2-3 MB
80	1.5-2.5 MB
70	1-1.5 MB
60	0.8-1.2 MB

The top of the scale is where the savings live, and the bottom is where the damage shows.

Why two photos at quality 80 have different sizes

Two photos exported at quality 80 can differ threefold, because JPEG compresses detail and photos hold different amounts of it. A clear sky compresses to almost nothing. A forest full of leaves barely compresses at all. Grain is the worst case: a high-ISO shot stays large at any quality setting, because the encoder treats the grain as detail worth keeping.

This is why a single recommended setting never works for every photo. A product shot on a white background and a concert photo in low light are different compression problems, and the same number treats them as if they were the same. Judge the setting on your own images, not on a chart.

JPEG quality vs file size in practice: the settings worth memorizing

75 to 85 covers nearly everything for photos on the web. Above 90 you pay for detail nobody can see. Below 60 you start seeing blocky patches in skies. Thumbnails can go lower, since nobody inspects a 200-pixel image. Archives stay at 90 or above, because storage is cheap and re-encoding a low-quality file later makes things worse.

Screenshots are the exception. Text and flat colors compress badly as JPEG, and quality 80 on a screenshot shows smudges around every letter. Save screenshots as PNG or WebP, and keep JPEG for photos.

How to pick a number instead of guessing

Export the same photo at 90, 80, and 70, then compare the three at 100 percent. Pick the lowest setting where you cannot spot a difference. That test takes two minutes and beats every rule of thumb, because it uses your photo instead of someone else's. When the limit is a number of kilobytes, a target-size tool earns its place.

I wrote a longer walkthrough of compressing JPEGs without visible loss if you want the full process with examples. When the limit is a number rather than a look, SnapShrink's exact-KB compressor aims at the number you type and finds the highest quality that fits under it. The slider is a fine tool for a look. A target size is the right tool for a number.

Questions about JPEG quality

Three questions come up whenever this topic shows up in a thread, so here are the short answers. The sections above cover the mechanism; these cover the edge cases people run into when they start changing quality settings on files they care about.

Does lowering JPEG quality damage the original?

Only if you overwrite it. Export to a new file and the original stays untouched. Work on a copy whenever the photo matters, because quality lost in a re-encode never comes back.

Is quality 100 the same as the original?

No. JPEG always discards some data, even at 100. The file is large and the loss is invisible, but it is still lossy. If you need every pixel preserved, use PNG or keep the RAW file.

Does resizing matter more than quality?

Often, yes. Halving the width and height removes three quarters of the pixels, and no quality setting can match that. Resize first to the size you need, then tune quality. Doing it the other way round wastes effort on pixels nobody will see.

The slider stops feeling random once you know the curve. High numbers buy detail you cannot see. Low numbers cost detail you can. Test your own photo at three settings, keep the lowest one that still looks right, and let a target-size tool handle the days when the limit is a number.