Nearly every image editing or export tool shows a quality slider labeled with a percentage, but that number is more abstract than it first appears — it isn't a direct measurement of "how close to the original" in any simple sense, and it doesn't mean the same thing across every tool.
What the percentage actually controls
A quality setting controls how aggressively the compression algorithm is allowed to simplify or discard image data — a higher percentage instructs the algorithm to preserve more detail (at a larger resulting file size), while a lower percentage permits more aggressive simplification (for a smaller file). It's a dial on the algorithm's behavior, not a direct percentage of "visual accuracy retained," even though it's often interpreted that way.
Why the same number means different things in different tools
Different image editors and export tools implement their JPEG or WebP quality scales slightly differently, so "80% quality" in one program isn't guaranteed to produce visually identical results, or even a similar file size, to "80% quality" in a different one. This is why comparing quality settings across different tools by their percentage number alone is unreliable — a visual side-by-side comparison of the actual output is more trustworthy than assuming the percentages behave identically.
Why quality loss isn't linear with the percentage
Dropping from 100% to 90% typically produces barely any visible change alongside a meaningful file-size reduction, because the algorithm is only trimming the least perceptually important data first. Dropping from 40% to 30% tends to show a much more dramatic visible difference for a comparatively smaller file-size gain, because by that point the algorithm is already cutting into data that matters more to how the image looks. This is why the useful range for most web images clusters around 70–85%, rather than being spread evenly across the full 0–100% scale.
A practical way to think about quality settings
Rather than treating the percentage as a precise, portable measurement, it's more useful to treat it as a per-tool, per-image dial: start around 80%, look at the actual result at real display size, and adjust up or down based on what you actually see rather than what the number implies. This approach works reliably across any tool, regardless of how that specific tool's quality scale happens to be calibrated internally.
Why some tools use a 1–100 scale and others use 0–1
Different tools and programming interfaces express the same underlying concept using different numeric scales — a percentage from 1 to 100, or a decimal from 0 to 1, or occasionally a different scale entirely depending on the specific compression library involved. These are just different units describing the same underlying quality dial, similar to measuring temperature in Celsius versus Fahrenheit, and converting between them (a decimal of 0.8 is equivalent to 80%) is usually straightforward once you know which scale a specific tool is using.
Try the image tool yourself — free, instant, no upload to any server.
Open the toolFrequently asked questions
Does 80% quality mean the image keeps 80% of its original data?
Not precisely — it's a setting that controls how aggressively the compression algorithm simplifies the image, not a literal percentage of retained data. The relationship between the number and actual visual or data difference isn't linear or exact.
Will 70% quality look the same in every image editing tool?
Not necessarily. Different tools calibrate their quality scales somewhat differently, so the same percentage can produce different file sizes or visual results depending on which specific tool is being used.
Why does dropping quality from 40% to 30% look worse than dropping from 90% to 80%?
Because compression algorithms remove the least noticeable data first. At higher quality settings, the data being trimmed is genuinely less important visually; at lower settings, the algorithm has already used up those "easy" savings and starts cutting into data that affects appearance more noticeably. Getting comfortable with this per-tool, visual approach tends to matter more in practice than memorizing any specific percentage as a universal rule to apply blindly across every situation. The same numeric setting can behave quite differently depending on the specific tool and the specific image involved, which is exactly why a visual check beats trusting the number alone.
