How to Compress Images Without Visibly Losing Quality
6 min read · Published October 14, 2026 · Updated October 14, 2026
You have a photo that is 6 MB, and the form you are filling in rejects anything over 500 KB. Or your website feels slow and the culprit turns out to be a handful of oversized pictures. In both cases the fix is the same: make the file smaller while keeping it good enough that nobody notices what you removed.
That is very doable, but only if you know which knobs matter. Most people reach for a quality slider first, when the biggest savings usually come from somewhere else. This guide walks through the options in the order that tends to work best, and explains when to stop.
Lossy and lossless compression are different bargains
Lossless compression shrinks a file by encoding the same information more efficiently, the way a zip archive does. When you open the result, every pixel is exactly what it was before. PNG works this way. The catch is that the savings are limited, especially for photographs, which contain a lot of fine, noisy detail that does not pack down well.
Lossy compression throws away information that people are unlikely to notice. JPEG and WebP (in its lossy mode) analyze the image and discard subtle color and texture differences. The file can shrink to a tenth of its original size or less, but the discarded data is gone for good. Saving a lossy file again and again, a habit called generation loss, adds a little more damage each time.
The practical rule: use lossless for things with sharp edges and flat colors, such as screenshots, logos and diagrams. Use lossy for photographs. And always keep your original file somewhere, so you can start over if you pick a setting that turns out too aggressive.
Resize first, then compress
File size grows with the number of pixels. A phone photo is often 4000 pixels wide or more, but a typical laptop screen shows a web image at well under 2000 pixels, and a page layout may display it at 800 or less. Every pixel beyond what will actually be displayed is dead weight.
Halving both the width and height removes three quarters of the pixels. Going from 4000 pixels wide to 2000 can cut the file size dramatically before you touch any quality setting, and the picture will look sharp at its intended size because nothing was blurred by heavy compression.
A quick way to choose dimensions is to look at where the image will be used. A blog column might be 700 to 900 pixels wide. A full-width hero image might be 1600 to 2000. An avatar might be 200. If you want sharp results on high-density screens, aim for roughly twice the displayed width, and no more.
- Profile photo or thumbnail: 200 to 400 pixels wide
- Image inside an article: 800 to 1200 pixels wide
- Full-width banner: 1600 to 2000 pixels wide
- Document scan for email: around 1600 pixels on the long edge is usually readable
Using a quality slider sensibly
For JPEG and lossy WebP, the quality setting is a number from 1 to 100, though the scale is not linear and differs between software. As a rule of thumb, values around 75 to 85 are often indistinguishable from the original at normal viewing size, while saving a large fraction of the bytes. Going from 100 to 90 usually saves far more than going from 90 to 80.
Below roughly 60, problems begin to show: blocky patches in smooth areas such as skies, fuzzy halos around text and sharp edges, and banding in gradients. Where those artifacts appear depends on the image. A busy forest photo hides compression well. A photo of a white wall with a shadow gradient does not.
ELDOSA's image compressor offers a quality slider from 10% to 100% for JPG and WebP, and it keeps the file in its original format. For PNG files the slider does not apply, because PNG is lossless in this tool; if a PNG is too large, resizing it or converting a photographic PNG to JPG or WebP will do more than any slider could.
Example: one 4000 x 3000 phone photo (illustrative)
Original JPEG at high quality -> several MB
Resize to 1600 px wide -> a few hundred KB
Then quality around 80 -> often a bit smaller still
Sharpness at 1600 px display: unchanged to the eyeChoose the right format before squeezing
Sometimes the best compression is switching formats. A photograph saved as PNG can be several times larger than the same photograph saved as JPEG or WebP, with no visible benefit. On the other hand, a screenshot of text saved as JPEG gets smudgy around the letters and may not even be smaller.
WebP usually produces smaller files than JPEG at similar visual quality and is supported by current mainstream browsers, though some older software and a few upload forms still insist on JPG or PNG. Check what the destination accepts before converting. The image converter on ELDOSA handles JPG, PNG and WebP in both directions.
Hitting a specific size such as 100 KB or 200 KB
Government portals, job applications and school systems often set a maximum such as 50, 100 or 200 KB. Meeting a limit by trial and error is tedious: you lower the quality, check the size, repeat.
The reliable approach is to resize first to a sensible dimension, then lower quality in steps until you land under the limit. If the picture is already down to around 60 to 70 quality and still too big, shrink the dimensions further rather than pushing quality lower, because small dimensions with decent quality usually look better than large dimensions with heavy artifacts.
ELDOSA has a target-size tool that automates this loop. You enter a size such as 100 KB, and it adjusts JPEG quality and dimensions until the result fits at or below that number. It does not promise the exact byte count, only that it stays under your limit. Be aware that the output is JPEG, so transparent areas are not preserved.
Knowing when to stop
There is a point where more compression is not worth it. Save the file, then view it at the size it will actually be shown. Zoom in to 100 percent and look at three places: smooth gradients, edges next to text or high-contrast lines, and faces. If none of those look wrong, you are done.
A useful habit is to compare the result side by side with the original. If you cannot tell them apart without magnifying, the extra savings from further compression are usually small next to the risk of visible damage. Also remember that an image going to print needs far more pixels than one going to a screen, so do not shrink print files with web settings.
- Keep the original and work on a copy
- Resize to the displayed size, then compress
- Stay near quality 75 to 85 for photos unless you must meet a limit
- Do not re-compress an already compressed JPEG repeatedly
- Check the result at 100 percent zoom before uploading
Tools mentioned in this guide
Frequently asked questions
Does compressing an image reduce its dimensions?
No. Compression changes how the pixel data is stored, not how many pixels there are. A 3000 pixel wide image stays 3000 pixels wide unless you resize it. Resizing and compressing are separate steps, and doing both usually gives the smallest file.
Why did my PNG barely get smaller?
PNG is lossless, so there is little to discard, especially for photographs. To shrink it meaningfully, reduce its dimensions or convert photographic content to JPG or WebP. Keep PNG for screenshots, logos and graphics that need transparency or crisp edges.
Is it safe to compress private photos online?
It depends on the service, since many upload your files to a server. ELDOSA's image tools process the file in your browser, so the image is not sent anywhere for compression. Whatever tool you use, read its privacy statement before uploading sensitive pictures.