How to Reduce JPG File Size Without Losing Quality

Why a JPG gets so heavy

JPEG is a lossy format. It throws away fine detail the human eye barely registers in order to make the file smaller. So a JPG should be light by definition, right? Then where do 5–10 MB photos come from?

There are usually three reasons, and they stack:

• Quality set too high. Cameras and editors save JPEG at quality 90–100 by default. That's "honest," but excessive: the difference between 80 and 100 is invisible to the eye, yet the file is 2–3 times heavier.
• Huge resolution. A modern phone shoots a 4000×3000-pixel frame or larger — 12+ megapixels. If that photo is then shown in a post 1080 px wide, you're carrying three quarters of the pixels for nothing.
• Extra data inside. EXIF (camera model, settings, GPS coordinates), color profiles, an embedded thumbnail — all of it lives inside the file and piles on kilobytes.

Quality 60–80: the main lever

JPEG has a quality parameter — a number from 1 to 100 that controls how aggressively the format discards detail. And here's the key fact that saves the most: quality 100 is almost never needed.

For the web and ordinary photos, the sweet spot is quality 60–80. A shot saved at 80 is indistinguishable from the original by eye, yet it weighs 2–3 times less. The reason is simple: the top quality values spend an enormous amount of data on differences that sit below the threshold of human perception.

Where's the line? Below quality 60, gradients (sky, skin, smooth transitions) and the area around sharp edges sprout the telltale "blocks" and halos — block-compression artifacts. That's the very "quality loss" everyone fears. So the practical rule is:

• 85–80 — when every detail matters: portfolios, small prints, text-heavy previews.
• 80–70 — the universal range for websites and social media. The best size-to-quality balance.
• 70–60 — backgrounds and secondary images, where saving bandwidth beats perfection.

Resizing: the most underrated trick

If you take one tip from this whole article, make it this one. Reducing the resolution to match the real display size shrinks the file more than anything else — by 80–90%.

The logic is airtight. A JPEG's size grows in proportion to the number of pixels. A 4000×3000 frame is 12 million pixels. If the image is displayed on the page at 1200 px wide (that's 1200×900 ≈ a little over a million pixels), you're storing and shipping ten times more data than necessary. Nobody will ever see those extra pixels — the screen downscales them anyway.

What dimensions should you target? Base it on where the image actually lives:

The rule is simple: there's no point storing an image wider than the slot it appears in (allow ×2 for high-density displays). Everything beyond that is dead weight.

Compress online in 30 seconds

If you need to shrink a few photos right now without fiddling with software, the fastest route is an online converter. It runs in the browser, there's nothing to install, and you can switch JPEG to a more efficient format in the same step.

If you need a lossless format instead — for graphics or text-heavy screenshots, say — the JPG to PNG converter has you covered. And the full list of conversions is always on the all formats page.

mozjpeg and jpegoptim for your own files

When you need to process hundreds of photos or bake compression into a site build, command-line tools take over. They work in batches, repeatably, and squeeze the most out of JPEG.

mozjpeg is an improved JPEG encoder from Mozilla. At the same visual quality it produces files roughly 5–15% smaller than the standard encoder, thanks to trellis quantization and smart progressive coding. It encodes more slowly than usual, but for batch work that doesn't matter. The command:

cjpeg -quality 80 -optimize -progressive input.jpg > output.jpg

Here -quality 80 sets the quality level, -optimize builds optimal Huffman tables, and -progressive makes the file progressive (more on that below).

jpegoptim is the tool when you want to squeeze already-finished JPEGs without re-encoding from scratch. It optimizes the Huffman tables (this is lossless) and can also enforce a quality ceiling. A few handy variants:

# Lossless: rebuild tables + strip metadata only
jpegoptim --strip-all photo.jpg

# With a quality ceiling of 80 and progressive output
jpegoptim --max=80 --all-progressive photo.jpg

# Squeeze under a size limit (e.g. 250 KB)
jpegoptim --size=250k photo.jpg

Progressive JPEG and stripping metadata

Two techniques that aren't the biggest win, but are essentially free and improve perception.

Progressive JPEG. A regular (baseline) JPEG loads top to bottom, line by line. A progressive one is encoded differently: it first shows a blurry version of the whole image, which gradually sharpens. The benefit is twofold — on a slow connection the user immediately sees what's loading, and the file is often a touch lighter than the baseline version. Turn it on with -progressive in cjpeg or --all-progressive in jpegoptim.

Stripping metadata. Inside almost every JPEG lives EXIF — technical info about the shot. There you'll find the camera model, shutter speed, ISO and, more importantly, often the GPS coordinates of where it was taken. This data:

• takes up space — tens of kilobytes, especially with an embedded thumbnail and color profile;
• leaks privacy — publishing a geotagged photo effectively announces where you are.

The command jpegoptim --strip-all photo.jpg removes the clutter in one move. For any photo headed to the internet, that's almost always the right call.

When WebP or AVIF beats JPEG

Sometimes the most effective way to "compress a JPG" is to stop using JPEG. The format is over 30 years old, and modern codecs beat it on efficiency by a wide margin.

At the same visual quality, WebP is roughly 25–35% lighter than JPEG, and AVIF about 50%. Both are supported in every current browser, so switching to them for a website is almost always a win for load speed and disk space.

When should you stay on JPEG? If you need absolute compatibility (old software, email clients, government portals) or the file format can't be changed by requirement. In every other case, for the web it's sensible to convert photos to WebP. For graphics, screenshots and images with transparency, PNG fits — but for photographs it's actually heavier than JPEG, so you'd choose it for reasons other than size.

Checklist: what to do right now

Short and ordered — from the most powerful move to the fine print:

Do the first two and, in most cases, the "JPG is too heavy" problem is solved. The rest is fine-tuning for those who want to squeeze out the maximum.