The Quality Slider Has Four Zones. Only One of Them Is Worth Using.

We ran the same source image through our converter twice, once left at the default setting, once dragged all the way down to 38%, deep into what our slider labels "Extreme Compression." The high setting came back at 65.61 KB. The 38% version came back at 33.86 KB, nearly half the size. Opened side by side, we couldn't spot a difference. It took zooming in to 380% before any gap in detail became visible at all.

That gap between "half the file size" and "you have to zoom in 380% to notice" is the entire subject of this article. For JPEG-style quality sliders, whether labeled 0-100 or "low/medium/high," this trade-off tends to fall into four practical zones, though the exact borders shift somewhat depending on the format, the specific encoder, and the image itself. Here's a useful rule of thumb for thinking about where those borders roughly sit.

Zone one (roughly 90-100% on most sliders): the zone that mostly wastes bytes

This is where people land out of caution, on the assumption that higher always means better. Above roughly 90%, that assumption stops paying off. Going from 85% up to 95% typically buys only a percent or two of measurable visual improvement while the file size can double or more. Research using the Structural Similarity Index, a metric built specifically to score images the way human perception actually works rather than measuring raw pixel differences, backs this up directly: studies looking for the point where viewers can no longer tell a compressed image from the original have consistently landed in the 92 to 96% SSIM range, well below what a 95%+ quality setting is spending your bytes to protect. This zone exists mainly for archival masters, print-bound files, or anywhere file size genuinely doesn't matter, not for anything headed to a screen.

Zone two (roughly 75-90% on most sliders): where almost everything belongs

This is the zone our own test landed in on the high setting, and it's the range that independent testing across formats keeps converging on as the actual sweet spot. Within this range, compression is discarding exactly the kind of information human vision is worst at noticing, subtle gradients and high-frequency detail, while preserving everything the eye actually locks onto: edges, faces, text, contrast. File size drops by roughly half to two-thirds compared to a maximum-quality export, with the resulting image staying visually indistinguishable at normal viewing distance. If you only remember one number from this entire article, it's that a setting somewhere in the high 70s to mid-80s is very likely where your image should sit.

Zone three (roughly 50-70% on most sliders): usable, but you're spending quality on purpose now

This is the zone our 38% test crept toward the edge of, and where our own side-by-side comparison only revealed a difference under heavy digital zoom. Below about 70%, you're no longer in "free lunch" territory, you're making a real trade, smaller files in exchange for artifacts that start becoming visible under close inspection, particularly on high-contrast edges like text sitting on top of a photo. For thumbnails, background images, or anything small on the page where nobody is zooming in, this range is genuinely fine. For a hero image or anything a visitor might actually examine closely, it's a riskier bet than zone two for a size saving that's often smaller than people expect.

Zone four (below roughly 50% on most sliders): past the point of no return

Below 50%, degradation stops being a subtlety and starts being obvious at normal viewing size: blockiness around edges, smearing in gradients, visible discoloration in areas that should be smooth. There's a narrow legitimate use for this zone, truly tiny thumbnails or placeholder previews where nobody is meant to study the image closely, but it's rarely the right choice for anything a visitor is meant to actually look at.

What actually happened in our own test

Here's the higher setting result, 36.36 KB source converted up to 65.61 KB.

And here's the same source dropped to 38%, landing at 33.86 KB instead, using the slider shown here.

Side by side on a normal screen, both looked identical. The difference only became visible once we zoomed to 380%, well past any zoom level a real visitor would ever use while browsing. Here's the higher-quality version at that zoom level.

And here's the 38% version at the same 380% zoom, next to it.

That's zone two versus the edge of zone three in practice: a real file size difference that a human being, looking at the image the way it's actually meant to be viewed, would never notice.

Finding your own border

These ranges are a starting point, not a fixed rule, since where the boundaries actually fall shifts by format and by image. The fastest way to find where your specific image's zones actually sit isn't a fixed number, it's a two-minute test: export the same image at roughly 85%, 70%, and 50%, then look at all three at the size they'll actually be displayed, not zoomed in. The setting where you stop being able to tell it apart from the original is your zone two, and it's usually higher than instinct suggests and lower than caution suggests. You can test your own images and quality settings using our converter directly in your browser.