How to compress an image without losing visible quality
You can usually remove 60-80% of a photo's file size before anyone notices. What ruins the picture is not compression itself, but lowering the wrong setting first.
Do them in this order
The order matters because each step has a different cost. Cropping and resizing discard pixels you were never going to display, so they cost nothing. Lowering quality re-encodes every pixel and eventually shows. People who start with the quality slider end up with a blurry image that could have stayed sharp.
- Crop: removing half the frame removes half the pixels and costs no quality at all
- Resize: halving the width removes three quarters of the pixels - the single biggest saving, and free
- Then lower quality: from 95% to 80% is invisible on most photographs and saves a lot
- Finally change format: WebP is 25-35% smaller than JPG at the same visible quality
What each setting actually costs
| Setting | Typical saving | Visible cost |
|---|---|---|
| Crop unused edges | Proportional to area removed | None |
| Halve the width | About 75% | None below ~2000 px for screen use |
| Quality 95% → 85% | 20-40% | None on most photographs |
| Quality 85% → 75% | 15-25% | Rarely visible, slight softening |
| Quality below 60% | Large | Artefacts around edges, faces, text |
| JPG → WebP | 25-35% | None visible at matched quality |
| JPG → AVIF | 25-50% | None visible at matched quality |
Where quality loss shows first
Lossy compression spends fewer bits where the eye is less sensitive, so damage appears in a predictable order: smooth gradients first (skies, skin, studio backgrounds), then fine texture, then edges, and last the overall shape. A photograph of a brick wall survives 60% quality; a portrait against a plain backdrop shows banding and blotches at the same setting, because the gradient is exactly what the encoder approximated.
Text and line art behave worst of all. If the image contains letters, screenshots of interfaces, or maps, stay in PNG or use a much higher quality threshold - or accept that the format is wrong for the content rather than the setting.
Does compressing twice make it worse?
Yes, and this is the mistake that surprises people most. Every lossy save re-encodes the whole image and adds its own damage on top of what is already there. Save the same photo as JPG at 80% five times and the quality declines each time even though the file size stops shrinking.
The fix is to keep a master - the original file, or a lossless export - and always generate the working copy from it. If you must work from a compressed file, minimise the number of saves rather than the number of edits: adjust everything in one session and save once.
Tools used in this guide
Compress Image
Shrink JPG, PNG and WebP images with quality, output-format and maximum-width controls. Batch several files at once, compare before/after sizes and download everything as a ZIP.
→Compress Image to Target Size
Shrink a JPG, PNG or WebP file to an exact size such as 100 KB, 200 KB or 1 MB. The quality is lowered step by step and, if necessary, the dimensions are reduced until the file fits the limit you need.
→Resize Image
Resize one or many images by exact pixels or by percentage while keeping the aspect ratio. High-quality downscaling, optional WebP or PNG output and batch ZIP download.
→Compress Image to 200KB
Compress an image to 200 KB in your browser. Job, visa and government portals often ask for exactly this - no upload, no signup, no watermark.
→Cite this page
Free to cite. Copy the reference below, or link to this page directly.
PicConv. "How to compress an image without losing visible quality". https://picconv.com/guides/compress-without-losing-quality/ (accessed 2026-11-01).
Frequently asked questions
What quality percentage is safe?
For most photographs 80-85% is indistinguishable from the original on screen. 75% is usually still fine. Below 60% artefacts become visible on close inspection, and below 40% they are obvious even in a thumbnail.
How much can I compress a photo without anyone noticing?
Typically 60-80% of the file size, if you crop and resize first. The most common mistake is leaving the dimensions at camera resolution while pushing quality down - reducing the pixel dimensions gives far more saving with far less visible damage.
Does compressing the same file twice reduce quality?
Yes. Each lossy save re-encodes and adds new damage to old damage. Keep the original master and re-export from it whenever you need a new size, rather than re-saving an already-compressed copy.
What about screenshots and text images?
Keep them in PNG. Lossy compression is built for photographic content; on text, edges and flat colour it produces blur and ringing artefacts while often not even saving much space.
Does compression remove metadata?
Re-encoding usually drops most EXIF data, including GPS coordinates. That is often desirable for privacy, but it also means the creation date and camera settings are gone. Remove metadata deliberately if you want certainty.