UtiliKit
Back to blog

JPEG Quality 80 vs 90 vs 100: A Side-by-Side With File Sizes

Published: November 13, 2026 - 7 min read

For photos on a website, JPEG quality 80 is the setting to use. In my test, going from 80 to 90 made the files 51% bigger with no difference I could see at normal viewing size, and going from 90 to 100 made them 2.75 times bigger again. Quality 100 isn't "the original". It's just a very large JPEG.

Below are the actual file sizes, what changes visually at each step, and the two cases where 90 is the better pick.

The test: four images, nine quality settings

I took four detailed images (two interiors with soft gradients on walls, two exteriors full of foliage and brick texture), resized each to 1600 px wide, and saved every one as a JPEG at quality 100, 95, 90, 85, 80, 75, 70, 60 and 50. Same encoder, standard 4:2:0 chroma subsampling, optimised tables. Here's the total for all four files at each setting.

QualityTotal size, 4 imagesCompared with quality 80
1004,887 KB4.2× larger
952,561 KB2.2× larger
901,776 KB51% larger
851,400 KB19% larger
801,175 KBbaseline
751,017 KB13% smaller
70915 KB22% smaller
60755 KB36% smaller
50653 KB44% smaller

Look at the shape of that. From 50 up to 80, each step costs a little. From 80 to 90 the price jumps, and from 90 to 100 it goes vertical. The scale isn't linear, and it isn't a percentage of anything. It's a dial that controls how coarsely the encoder rounds off fine detail, and the top of the dial is where rounding almost stops and the file balloons.

One single image makes the point better than a total. A 1600 by 900 interior came out at 85 KB at quality 80, 130 KB at 90 and 424 KB at 100. That's five times the bytes between 80 and 100, for the same picture.

What you can actually see at 80, 90 and 100

I compared the files side by side at 100% zoom, then at 300%.

At 100% zoom, on a normal laptop screen, I couldn't tell 80 from 90 from 100 on any of the four images. Not on the brick, not on the leaves, not on the smooth walls.

At 300% zoom there are differences, and they're in predictable places. At quality 80, thin dark lines against a bright sky (window frames, branches) pick up a faint halo a pixel or two wide, and very smooth gradients show the first hint of 8-pixel blocks if you know to look for them. At 90 those are gone for practical purposes. At 100 the image is as clean as JPEG gets, though it still isn't identical to the source, because JPEG at 100 still converts colour and still rounds.

Below 80 things change faster. At 70 the halos are visible at 200% zoom. At 60 flat areas such as painted walls start to look slightly patchy, and at 50 you can see blocking in skies at normal size on a good monitor. For a thumbnail, 60 is fine. For a full-width photo, I wouldn't go under 75.

So where's the visible gain for the extra 51% between 80 and 90? For an ordinary photo viewed at ordinary size, there isn't one.

Why quality 100 is rarely the right answer

People pick 100 because it sounds like "no loss". It isn't lossless. A JPEG at 100 has already thrown away half of the colour resolution in each direction (that's the chroma subsampling) and rounded every block. If you want a truly lossless file, you want PNG, and for photos that's bigger again: the four test images as PNG came to about 9.7 MB against 4.9 MB at JPEG 100 and 1.2 MB at JPEG 80.

There's a second trap. Re-saving an existing JPEG at a higher quality doesn't bring anything back. I took the 85 KB quality-80 interior and re-saved it at 100. The result was 236 KB, nearly three times the size, and it contained exactly the detail of the 85 KB file, halos included. The encoder was faithfully preserving the artefacts.

That happens a lot with images that have been through a phone app or a messaging service. They arrive already compressed, somebody "improves" them by exporting at 100, and the page gets heavier for nothing.

When 90 is worth the extra bytes

There are two honest cases for 90.

The first is an image that will be edited and saved again. Every lossy save costs a little, and although the loss from re-saving at the same setting is small (ten re-saves at 80 moved my test image by a barely measurable amount), cropping, resizing or colour changes between saves reset the damage each time. If the file is a working copy and not the final export, give it headroom: 90 or 95, or keep a PNG master and export JPEGs from that.

The second is an image with large, very smooth gradients or fine text. Product shots on a soft grey backdrop, a clear evening sky, a screenshot with small type. Those show banding and halos sooner than a busy photo does, because there's no texture to hide them in. For those I'd use 85 to 90 and check the gradient at 200% before publishing. For screenshots and graphics with text, JPEG is the wrong format in the first place, and a PNG or WebP will be both sharper and smaller.

Print is sometimes named as a third case. If a printer or a photo lab asks for maximum quality, give them 95 or more and don't worry about it, since nobody's waiting for that file to load on mobile data.

Pixels matter more than the quality number

This is the part that gets missed. The quality setting is the second-biggest lever on file size, not the first. Pixel dimensions are first.

A 4000 px wide phone photo saved at quality 80 is still around 1.5 to 2.5 MB. The same photo resized to 1600 px wide at quality 80 is typically 200 to 400 KB, and it looks identical in a page column that's 800 px wide. Dropping that 4000 px original from 90 to 60 saves less than resizing it does, and it looks worse. So the order of work is: resize to what the page really needs, then pick the quality.

The Image Resizer handles the first step. A sensible target is twice the displayed width for sharp results on high-density screens, so 1600 px for an 800 px column.

Format is the third lever. The same four images as WebP at quality 80 came to 878 KB against 1,175 KB for JPEG 80, about 25% smaller overall (between 15% and 51% depending on the image), and every current browser displays WebP. If your platform accepts it, that's free. If it only takes JPG, 80 is still the answer.

Short answer: which setting to use

One caution. Quality numbers aren't identical between programs. Photoshop's old 0 to 12 scale, a phone's "high" and a browser tool's 80 don't map onto each other exactly, so treat 80 as a starting point and look at the result. If the file has to fit under a hard limit, such as 200 KB for an upload form, it's easier to let the tool search for the setting: Compress to a File Size does that and reports the size of the encoded file, so you're not guessing at numbers.

To try the comparison on your own photo, open it in the Image Compressor, move the quality slider between 80 and 90, and watch the size readout while you look at the preview. It runs in your browser, so the photo isn't uploaded anywhere. If you can't see the difference, take the smaller file.

Frequently asked questions

Is JPEG quality 100 lossless?

No. JPEG at 100 still reduces colour resolution and rounds pixel values, so it never matches the source exactly. It only minimises the loss, at a large cost in file size. For a truly lossless copy use PNG, or lossless WebP.

Does re-saving a JPEG at a higher quality improve it?

No. Detail removed by an earlier save is gone, and a higher setting only stores the existing artefacts more precisely. In my test an 85 KB file re-saved at 100 grew to 236 KB with no visible change.

What JPEG quality do phones and cameras use?

It varies by maker and mode, but camera JPEGs are generally saved at a high setting, roughly equivalent to the 90s, which is why a 12-megapixel photo is often 3 to 5 MB. That leaves room to resize and re-export at 80 for the web.

Is quality 80 in one app the same as 80 in another?

Not exactly. Encoders use different tables and some switch colour subsampling at certain levels, so two files saved at 80 can differ in size by 10 to 20 percent. Use the number as a starting point and judge by the preview and the resulting file size.

More guides