How to Reduce Image Size Without Losing Quality

How to reduce image size without losing quality: the truly lossless methods, the visually lossless settings, and the real limits, with checkable numbers.

By Mike Wiberg · · 16 min read

How to reduce image size without losing quality comes down to one decision: whether you need bit for bit identical pixels, or pixels that only need to look identical. Truly lossless methods, stripping metadata and re-encoding in an efficient format, shrink a file with zero data loss. Visually lossless settings discard detail your eye cannot find, and that is where the big savings live.

I built the compressor on this site, so the numbers on this page come from the real tool and from the format specifications, not from a landing-page slogan. The general walkthrough of every route lives in the complete image compression guide. This page is about a single promise you have probably typed into a search box, compress image without losing quality: a smaller file that looks exactly the same, where that promise holds and where it quietly turns into marketing copy.

Here is the decision up front. Each row is expanded below.

What you actually need The route to take
Zero data loss, every pixel intact Strip metadata, re-encode losslessly, pick the smaller lossless format
A photo that looks the same but is far smaller Quality 80, or convert to WebP or AVIF
The image only shown at a fixed width Resize to that width first, then compress
Text, a logo, or a screenshot Keep it lossless, never save it as a lossy JPEG
A hard limit like 100 KB Combine resize and quality, checking the number as you go

What does "without losing quality" really mean?

It means one of two different things, and every tool making the promise is quietly using one of them. Knowing which one is the difference between a realistic expectation and disappointment.

Lossless: every pixel, fewer bytes

Lossless compression stores the exact same pixels in fewer bytes, usually by recording each pixel as a small difference from its neighbors instead of spelling every value out in full. Decode the file and you get back every pixel, identical to the original. PNG works this way, and WebP and AVIF both offer lossless modes. Google's web.dev performance guide describes the mechanism the same way I see it in the encoder, pixels packed more efficiently with nothing thrown away.

The catch is the ceiling. A lossless pass trims a file, it does not transform it. You will not turn a 4.5 MB phone photo into a 200 KB upload without discarding something, no matter how clever the packing.

Visually lossless compression

Visually lossless is the second meaning, and it is the one behind most large before and after numbers. The compression is lossy, data really is discarded, but the settings are conservative enough that you cannot spot the difference at normal viewing size. The quality slider is the dial. In ImageOptim's JPEG optimizer, a setting of 100 makes the pass lossless, and anything below 100 trades data for bytes (ImageOptim preferences). The default of 80 on our compressor sits in the visually lossless zone for photographs, which is why the output looks like the input while the file is a fraction of the size.

The two roads have different ceilings, and that is the whole point. Lossless keeps the pixels and takes what packing gains it can. Visually lossless spends detail your eye was never going to resolve and gets far more back. The full mechanics of both are in lossy versus lossless, explained.

Which methods keep every pixel?

Three methods are truly lossless, they change no pixel at all, and they stack on top of each other. Do all three to a screenshot and you have saved real bytes without touching the image.

Strip the metadata

EXIF data, the hidden block your camera writes into every file, carries GPS coordinates, the camera model, the timestamp, and a small embedded thumbnail. It adds bytes and shows nothing on screen, because it sits alongside the pixels rather than inside them. Removing it shrinks the file and improves your privacy in the same move. We strip EXIF by default, so this happens without you hunting for a checkbox.

Lossless image compression and smaller lossless formats

The same pixels can be packed well or badly, and plenty of files in the wild are packed badly. Lossless image compression re-encodes them into a tighter container with every pixel intact. PNG compression is lossless, so a PNG run through a good optimizer comes back identical and smaller. Better still, switch container: lossless WebP is typically around 26 percent smaller than the same image saved as PNG, with nothing lost (MDN on image file types). And when an image uses only a handful of colors, a logo or a flat chart, indexed PNG-8 stores it losslessly in a fraction of the bytes of full-color PNG, because it only records which of a few colors each pixel uses.

Notice what none of these can do: they cannot make a photograph small enough for a tight upload limit. For that you need the second road.

How to reduce image size without losing quality, step by step

The fastest way to reduce image size without losing quality on a photo is a browser compressor at the default quality, because there is nothing to install and the setting is already in the visually lossless zone. I use the one on this site here because I built it and can vouch for its numbers byte for byte.

  1. Open the free compressor here and drop your image on it. You can drop up to 30 images at once, up to 25 MB each, with no signup.
  2. Leave the quality at 80. That is the level the web has settled on as visually lossless for photographs.
  3. Read the after size on the result card. If you need the file smaller, pull the quality slider down and watch the number fall until it fits. Re-compressing the same upload at a new quality or format never counts against you.
  4. If the file is a screenshot, a logo, or line art, choose PNG or lossless WebP on output instead of dropping quality. Lossy settings smear sharp edges, and this is the content they hurt most.
  5. Download the result. The original on your device is untouched, and the uploaded copy is deleted from our storage within the hour.

Two useful things happen on the way through that you would otherwise do by hand: EXIF is stripped, and JPEG output is written progressive, so on a slow connection it renders in passes instead of loading top to bottom.

How much can you shrink a photo before the eye notices?

A photograph off a phone camera typically comes back 60 to 80 percent smaller at quality 80, with no difference you can see at normal size. The consistent example from our own tool is a 4.5 MB phone photo landing around 900 KB. That is a visually lossless result, not a lossless one, and it works because lossy encoders discard exactly the data human eyes are worst at seeing. They round fine detail to coarser values, a step called quantization, and they store color at lower resolution than brightness, called chroma subsampling, because your eye reads brightness far more sharply than color (web.dev on image performance).

Format choice adds a second multiplier

On top of the quality setting, the container matters. WebP is typically 25 to 35 percent smaller than a JPEG of the same visual quality, and AVIF is often around 50 percent smaller than JPEG, at the cost of slower encoding. The demo file on our home page is the checkable case: a 497 KB JPG comes out as a 214 KB WebP at quality 80, which is 56.9 percent smaller with the same picture on screen. Stacking a format switch onto a generous source file is how the headline "up to 88 percent smaller" happens.

The table and chart below are the honest savings menu, read from smallest cut to largest. The pattern is the point: the less attached you are to bit-perfect pixels, the more bytes you get back.

Route Typical size reduction Visible quality cost
Lossless WebP instead of PNG About 26 percent None, pixels identical
Lossy WebP instead of JPEG 25 to 35 percent None at matched quality
AVIF instead of JPEG About 50 percent None visible, slower to encode
Quality 80 on a phone photo 60 to 80 percent Invisible at normal size
Size cut with no visible quality loss percent smaller, by route, same picture on screen Lossless WebP vs PNG about 26% WebP vs JPEG 25 to 35% AVIF vs JPEG about 50% Quality 80, phone photo Figures from imagecompressor.tools and the format specs. Bar length proportional to percent saved.

If the photo is going somewhere you do not control, one caveat: AVIF support is still not universal, so WebP is the safer modern default and a JPEG copy remains the universal fallback. The format tradeoffs in full are in WebP and AVIF, the modern formats.

Built-in ways on Windows, Mac, and your phone

You do not always need a website. Every desktop and phone ships a tool that reduces file size, though most of them resize rather than re-encode, so the quality tradeoff is different. Reach for these when the file will not leave your machine or you have no connection.

Windows

Open the image in the Photos app, use the menu at the top, and choose Resize. Photos lets you drop the dimensions or the quality, which shrinks the file. This is resampling, so you are trading pixels, not just repacking them. Keep a copy of the original before you overwrite it.

Mac

In Finder, right-click the image and pick Quick Actions, then Convert Image, and choose a smaller size. Or open it in Preview, use Tools then Adjust Size, and drop the dimensions. One detail worth knowing: in Preview's Adjust Size dialog, unticking Resample changes the print dimensions without touching the pixel count, so you can re-label a photo for printing without discarding anything (Apple Support on resizing in Preview).

Phones

On a phone, the browser route is the simplest honest answer: the compressor here works the same in mobile Safari or Chrome, no app to install. The native photo editors can crop and, on some models, export a smaller copy, but the exact controls differ by model and version, so I cover them where I can test them, in compress images on an iPhone and compress images on Android.

Photoshop and other pro editors

Professional editors give you the same two levers, quality and dimensions, with finer control. In Photoshop, Export As or the older Save for Web lets you preview the file at each quality level and stop at the last setting that still looks right, which is the manual version of what an automatic quality algorithm does (Cloudinary on image optimization). When you downsize, choose the resampling method built for reduction so edges stay crisp. The point is not the brand of editor, it is that you experiment with the compression level until quality and size meet, rather than trusting a single preset. The step-by-step is in compress an image in Photoshop.

Is resizing the same as compressing?

No, and confusing the two is the most common way people lose quality they meant to keep. Resampling removes pixels: a 4000 pixel wide photo resized to 1200 pixels has permanently thrown away most of its pixel data. Re-encoding keeps every pixel position and changes only how the values are stored. Both make the file smaller, but only one changes the resolution.

Resizing is still often the right call. If an image will only ever be shown at 1200 pixels wide, the extra pixels are dead weight, and dropping them costs nothing you will ever see in that context. The trap is doing it to your only copy. Once pixels are gone, no setting brings them back, because upscalers interpolate, meaning they guess plausible pixels rather than recover real ones. The honest state of that art is in can you increase image resolution.

What if you need an exact size, like 100 KB?

When a form demands a hard number, combine the two levers in order: resize first, then lower quality. Dimensions buy the large reductions, because file size scales roughly with pixel count, so halving the width and height cuts the raw pixel budget to a quarter. Quality then buys the final factor of two. A worked case makes the order obvious: a 3000 pixel wide photo squeezed down to a 500 pixel avatar keeps only about three percent of its original pixels, so a tight 100 KB limit is usually met on the resize alone, before you touch the quality slider at all. Drop the photo in, resize to the largest dimensions the destination actually displays, then pull the quality slider down while watching the after number until it clears the limit with a little room to spare.

This is a deliberate trade, not a free one, so make it once from the best source you have and check the result by eye before you ship it. The exact-target workflow, with the settings that hit common limits, is in reduce image file size in KB. If the file is bound for an upload form specifically, compress images for uploads covers the limits the big destinations enforce.

What actually ruins image quality?

One careful compression pass at a sane setting ruins nothing. The visible damage in real life comes from four repeat offenders, and all four are avoidable.

Repeated lossy saves

Every lossy save discards a little more data, and the losses stack, generation after generation. This is generation loss, and it is why a photo edited and re-saved five times looks soft and blotchy while a photo compressed once looks fine. Compress once, from the best source you have, as the last step before the file leaves your hands.

Lossy compression on the wrong content

Text, line art, charts, and sharp-edged logos are exactly what quantization smears. Screenshots of interfaces and diagrams belong in a lossless format; photographs belong in a lossy one. One rule, most of the benefit.

Platforms that recompress after you

Messaging apps, social networks, and photo services re-encode uploads on their own servers, at settings you do not choose and mostly cannot see. If a photo looks worse after it went somewhere, that somewhere probably recompressed it. What the big apps actually do to uploaded files, measured with real byte counts, is the subject of which apps compress your images.

Editing the compressed copy instead of the original

Crop and adjust the original, keep it somewhere safe, and export fresh copies from it whenever a new size is needed. Editing a already-compressed file just bakes in the losses. For what a single pass does and does not cost you, see does compressing an image reduce quality.

Where to go next

Each piece of this topic has its own full guide in this cluster and the two sibling pillars pick up the routes this page only summarizes.

Questions

Can you really compress an image with zero quality loss?

Yes, within limits. Lossless compression repacks the same pixels into fewer bytes, so nothing is discarded, and stripping metadata and re-encoding a PNG are both lossless. The catch is the ceiling: lossless passes trim a file, they do not transform it. The large 60 to 80 percent reductions you see on photographs come from visually lossless settings, which discard data your eye cannot resolve.

What quality setting avoids visible loss?

Use 80 for photographs. It is the default on our compressor because photographic detail survives at normal viewing size while the file drops 60 to 80 percent. A setting of 100 in a lossless-capable optimizer keeps every bit. If an upload limit forces you well below 80, check the result by eye before shipping it, because every step down discards more data.

How do I reduce image size without losing quality on Windows?

Open the image in the Photos app, use the top menu, and choose Resize, then lower the dimensions or quality. That method resamples, so it trades pixels for bytes rather than repacking them; keep the original first. For a repack with no pixel loss, run the file through a browser compressor and save it as lossless WebP or optimized PNG instead.

Does converting a JPEG to PNG restore quality?

No. Whatever the JPEG encoder discarded is gone, and saving those pixels losslessly afterward just preserves the damage in a much larger file. A photo converted from JPEG to PNG usually grows dramatically while looking identical, artifacts included. Work the other way: keep one high quality original, and export compressed copies from it whenever you need a new size or format.

Why does my photo look worse every time I save it?

Because every lossy save discards a little more data, and the losses stack. This is generation loss, and several rounds of editing and re-saving can visibly degrade a photo even at a high quality setting. The fix is to keep one original, make all edits on that file, and export a compressed copy once, as the final step before the file leaves your hands. The short version of this whole page: decide first whether you need bit-perfect pixels or pixels that look perfect, because the second is far cheaper. When you have a photo that needs to shrink right now, [drop it on the compressor and compare the before and after yourself](/). No signup, up to 25 MB per image and 30 at a time, and the uploaded copy is deleted within the hour.