Halving an image's width and height removes three quarters of its pixels, not half. That squared relationship is why resizing beats every compression setting.
File size is driven, before compression, by how many pixels there are. And pixel count is a product of two dimensions, so changing them both compounds.
People consistently underestimate this. "I only halved it" feels like a 50% saving and is actually 75%. It is the single most useful piece of arithmetic in the whole subject.
Uncompressed, file size tracks pixel count exactly. After compression it tracks it closely but not perfectly, for two reasons.
First, headers and metadata are a fixed overhead, which dominates on very small files — this is why a 100 × 100 thumbnail is not one four-hundredth of a 2000 × 2000 image.
Second, downscaling averages neighbouring pixels together, which removes fine noise. Less noise means less unpredictable detail for the encoder to describe, so a downscaled image often compresses better per pixel than the original did. In practice this means resizing frequently beats the arithmetic rather than merely matching it.
Applied in that order to a typical 4.8 MB camera photograph headed for an 800-pixel column: resize to 1600 px (about 900 KB), convert to WebP at quality 78 (about 130 KB). A 97% reduction, and it looks identical where it is displayed.
Already-compressed sources. A file that has been through a good encoder has little redundancy left, so further compression achieves little. Resizing still works, because it removes pixels rather than redundancy.
Highly detailed content. Foliage, gravel, crowds and heavy grain need genuinely more bytes per pixel than a portrait against a plain wall. Two images at identical dimensions can differ threefold in size, legitimately.
Lossless formats. PNG ignores quality entirely, so resolution is the only lever you have — which is exactly why oversized PNG screenshots are such a common problem.
Everything below runs in your browser, so you can test any of this on your own files without uploading them.
By pixels or percentage, aspect ratio locked, across a whole batch.
Open the toolShrink JPG and JPEG photographs with a quality slider or an exact target size.
Open the toolTypically 25–35% smaller at matched quality. The standard web win.
Open the toolFull resolution kept, quality still invisible to the eye.
Open the toolQuestions
Every answer below is present in the page source, expanded, so it can be read without opening anything.
Usually by a wide margin. Halving both dimensions removes three quarters of the pixels, which typically cuts the file by a similar proportion — more than you can safely gain from the quality slider, and without visible cost if the image is still larger than its display size.
No — it drops to roughly a quarter, because you have halved both width and height. This is the point people most often miss, and it is why resizing is such an effective first step.
Fixed overhead. Headers, colour profiles and metadata occupy a constant number of bytes regardless of dimensions, so they dominate on very small files. A 100 × 100 thumbnail cannot be one four-hundredth of a 2000 × 2000 image.
Resize. Compressing first spends quality on pixels you are about to discard, and the result is worse than doing it the other way round for the same final size.
More so, because PNG is lossless and has no quality setting at all. Reducing dimensions is essentially the only lever available for a PNG, short of changing format.
Content. Detailed subjects — foliage, texture, crowds, grain — contain genuine fine detail with no repetition, and the encoder has to pay for it. A portrait against a plain wall compresses several times better at identical dimensions.
From the blog
Longer, more practical guides from the blog.
Resize before you compress. Measured on one photograph: half the width at quality 80 produced a 16.4 KB file, while full width at quality 40 produced 20.6 KB and looked far worse. The order of operations matters more than the settings.
8 min read
Quality 80 is not 80% of anything. It is an index into a quantisation table. Measured on one photograph: dropping from 100 to 90 removes 77% of the file and 4.2 dB of fidelity; dropping from 40 to 30 removes 17% and costs almost as much.
8 min read
PNG stored the same photograph in 1274.1 KB where JPEG needed 67.5 KB — nineteen times larger. On a flat graphic PNG won. On a smooth gradient PNG lost to JPEG by a factor of two, which surprises most people.
9 min read