How to reduce image file size
The upload form wants under 2 MB and my photo is 8 MB. What do I actually change?
Three levers, and one of them dominates
Every image compressor is pulling some combination of three levers: how many pixels the image has, how much detail the encoder is allowed to discard, and which format it writes. People reach for the quality slider first because it is the one every tool shows first. It is usually the wrong place to start.
Dimensions dominate, because file size scales roughly with pixel count and pixel count scales with the square of the linear size. Halving the width and the height leaves a quarter of the pixels. Nothing you can do with a quality slider is that decisive, and nothing else costs so little in perceived quality — provided the result is still larger than the box it will be displayed in.
Work out how many pixels you actually need
A modern phone camera produces images in the region of 4000 pixels wide. A full-screen photo on a laptop is displayed at something like 1500 to 2000 pixels wide, and a photo inside an article is often displayed at 800. Sending 4000 pixels to a box that shows 800 means about 96 per cent of the pixels are discarded by the viewer's software after they have been downloaded.
| Longest edge | Approximate pixels | Share of the 4000 px original |
|---|---|---|
| 4000 px | 10.7 million | 100% |
| 2560 px | 4.4 million | 41% |
| 1920 px | 2.5 million | 23% |
| 1280 px | 1.1 million | 10% |
| 800 px | 0.4 million | 4% |
Those percentages are arithmetic, not a measurement of any particular file — a compressed file does not shrink in exact proportion to its pixel count. They are still the right thing to look at first, because they tell you how much room the resize is giving the encoder before it has to start discarding detail.
What the quality slider is doing
In a JPEG encoder, the quality number scales a table of divisors applied to the frequency coefficients of each 8×8 block. Lower quality means coarser division, which means more coefficients round to zero, which means fewer bits. The visible cost appears first around hard edges and in flat gradients: ringing next to text, and banding in skies.
The scale is not linear and it is not comparable across formats. Quality 80 in a JPEG encoder, a WebP encoder and an AVIF encoder are three unrelated numbers that happen to share a range. Judge the result by looking at it at full size, not by trusting the number.
Why a target size is a search, not a setting
No encoder accepts "make this 500 KB" as an instruction. It accepts a quality setting and produces whatever size that setting produces for that particular image. A tool that offers you a target size is therefore running a search: encode, measure, adjust the quality, encode again, and stop when the result is under the budget or when the quality floor is reached.
That is why a target is always best-effort. A detailed, noisy, high-resolution photograph may not fit a tight budget at any quality a person would accept, and the honest response is to stop at the floor and tell you where it stopped rather than to keep going until the picture is ruined. FileSlimmer's target-size tools reduce dimensions first, then narrow quality between bounds, and report the setting they finished on.
How to make a picture smaller than 1 MB
Start by resizing a phone photo to the largest dimensions the destination needs. For email, forms and ordinary web pages, that is often far below the camera original. Then encode a fresh JPEG or WebP at a high-but-not-maximum quality, measure the result and lower quality gradually only if the file is still above 1 MB.
Quality alone does not determine whether a result lands below 1 MB, because detailed images need more data than simple ones. If the first pass misses the limit, reduce dimensions before forcing quality lower. This normally preserves a more natural-looking image than keeping every pixel and adding strong block or ringing artefacts.
Things that do not help
- Putting the file in a ZIP. Compressed image formats have almost no redundancy left for a general-purpose compressor to find; the archive is usually within a rounding error of the original size.
- Renaming the extension. The bytes do not change, and software that reads the actual file signature will simply be confused.
- Re-saving a JPEG at the same quality. It is a second lossy pass: the file gets no smaller in any useful way and the picture gets slightly worse.
- Stripping metadata from a photograph, if size is your only goal. A few kilobytes of EXIF against several megabytes of pixels is noise. Strip it for privacy reasons, which are real, not for size.
A workflow that usually gets there
- Decide the largest size the image will ever be displayed at, and resize the longest edge to that.
- Choose the format that fits the content: a lossy format for photographs, a lossless one for screenshots and flat graphics.
- Start at a high-but-not-maximum quality and step down until you can see the damage, then step back up once.
- Check the result at full size, on the kind of screen it is meant for.
- If it is still too large, go back to the first step. There is almost always more room in the dimensions than in the quality.
Where this stops working
Some images resist all of this. A scanned document full of fine text needs its pixels to stay legible, so resizing hurts immediately and quality reduction produces exactly the artefacts that make text hard to read. A screenshot of a spreadsheet has the same problem. For those, the right answer is often a different format — lossless, or a PDF — rather than a smaller number of kilobytes.
And if a portal rejects your file, check whether the limit is on the file size or on the pixel dimensions. They are different limits, and a file that passes one can still fail the other.
Tools for this
Sources
- ITU-T T.81 — quantisation and the JPEG quality scale
- MDN — Image file type and format guide
- MDN — Responsive images and choosing a display width