How to Compress JPEG Without Blur

You export a JPEG from Lightroom, Photoshop, or your phone at a high quality setting and it's 6 MB. You drop the quality slider to get it under 500 KB and the result looks smeared — fine textures gone, sharp edges replaced by ghostly halos, and blocky 8×8 squares visible across flat areas. The image is smaller, but it's now unusable.

The frustrating part: that blur isn't inevitable. It's what happens when the quality setting crosses the wrong threshold, or when the wrong encoder is used. With the right settings and tools, you can cut a JPEG's file size by 50–65% while keeping it visually sharp at any normal viewing size.

What is JPEG compression?

JPEG (Joint Photographic Experts Group) is a lossy image format — unlike PNG, it permanently discards some image data during compression. The way it discards data is what determines whether the result looks sharp or blurry.

JPEG compression works by dividing the image into 8×8 pixel blocks and applying a mathematical transform called DCT (Discrete Cosine Transform) to each one. This converts pixel values into frequency components — essentially separating the broad strokes of the image (low frequency) from fine detail and sharp edges (high frequency). At lower quality settings, the high-frequency components are discarded first. That's the detail that makes text crisp, fur texture visible, and foliage sharp. When it's gone, those areas turn into smooth, smeared blobs.

The key insight: there's a wide range of quality settings where the high-frequency detail is preserved but the file is still dramatically smaller than an unoptimized export. That range — roughly 75–92% depending on the image — is the target zone for blur-free compression.

• Quality 85–92% — visually lossless for almost all photographic content. Indistinguishable from the original at normal viewing sizes. Best for product photography, hero images, and print-ready exports.
• Quality 75–84% — imperceptible quality loss on most photographs at web viewing sizes. The sweet spot for web performance. 40–60% smaller than a 100% export.
• Quality 60–74% — noticeable to trained eyes on close inspection. Acceptable for low-resolution thumbnails, email previews, or social media where further re-compression will happen anyway.
• Below 60% — visible blocking and blur artifacts at normal viewing sizes. Avoid for any image the user is meant to actually look at.

Why it matters for your site and workflow

JPEG is the dominant format for photographic web content. Every product image, hero photo, blog illustration, and editorial image that renders in a browser is almost certainly a JPEG. How well those JPEGs are compressed directly determines the performance and perceived quality of your pages.

• Core Web Vitals. Google's LCP (Largest Contentful Paint) metric directly measures how quickly the main image in a viewport loads. An unoptimized 4 MB hero JPEG on a product page can add 3–5 seconds of load time on a mobile connection — pushing LCP well past the 2.5-second "good" threshold and impacting search rankings.
• Bandwidth and hosting costs. If your site serves 100,000 image requests per day, the difference between a 2 MB and a 700 KB JPEG is over 100 GB of bandwidth monthly. At CDN rates, that compounds fast.
• User perception of quality. Paradoxically, a well-compressed JPEG at 80% quality often looks sharper in the browser than an unoptimized export — because it loads faster and the browser doesn't have to decode and scale a massive file on-the-fly, introducing its own resampling blur.
• Email and social media. Most email clients and social platforms re-compress any JPEG you upload to their own quality settings. Sending a 5 MB original means the platform's compression algorithm decides the final quality — often aggressively. Sending a well-compressed version at 82% gives the algorithm less to destroy.
• E-commerce conversion. Product image load time has a direct, measurable impact on add-to-cart rates. A 1-second delay in image rendering is consistently associated with conversion drops of 7–12% across e-commerce studies.

Step-by-step: how to compress a JPEG without blur

1. Always start from the original source file. The single most important rule in JPEG compression is to never re-compress an already-compressed JPEG. Every round-trip through JPEG compression compounds artifacts — what engineers call "generation loss." Open the RAW file, the PSD, the Lightroom export at 100%, or the highest-quality source you have. Never work from a previously saved JPEG if you can avoid it.
2. Identify the image content type. Is it a photograph with complex color gradients and fine textures? A screenshot with text? An illustration? JPEG handles photographs well. For screenshots, graphics with text, or images with solid color areas and sharp edges, JPEG compression at any setting produces visible artifacts — those images should be PNG or WebP instead. JPEG compression quality is only worth tuning for genuine photographic content.
3. Choose your target quality range. For web use, start at 82%. For product photography that needs to hold up at larger display sizes, start at 88%. For thumbnails under 200px wide, 70% is usually sufficient. These are starting points — the right value depends on the specific image.
4. Use a MozJPEG-powered encoder, not standard libjpeg. MozJPEG is Mozilla's open-source JPEG encoder that produces 10–20% smaller files than the standard encoder at identical quality settings. It uses better quantization tables and trellis quantization. The Rebrixe JPEG Compressor, Squoosh, and Sharp (Node.js library) all use MozJPEG under the hood. If your current tool doesn't specify the encoder, it's almost certainly using the inferior standard encoder.
5. Enable progressive encoding. Progressive JPEGs encode the image in multiple low-to-high resolution passes instead of top-to-bottom. The result is typically 10–15% smaller than the equivalent baseline JPEG, and the user sees the whole image almost immediately (blurry, then sharpening) rather than watching it paint from top to bottom. Set this in your export options wherever available.
6. Strip metadata before export. Camera JPEGs contain embedded EXIF data — GPS coordinates, camera model, lens information, color profiles, and thumbnail previews. This can add 50–200 KB to a file that might otherwise be 300 KB. Strip it using ExifTool, Squoosh's "Remove EXIF" option, or the Rebrixe JPEG Compressor. Never skip this step for web images.
7. Preview at 100% zoom and compare side by side. Open the compressed output at 100% zoom alongside the original. Look at the sharpest areas of the image: text in the background, hair or fur, foliage, textured fabric. If you see haloing around edges or blocky 8×8 squares in smooth areas, raise the quality by 5 points and re-compress. If it looks identical, try dropping quality by 3 points to see if you can squeeze more savings without visible change.
8. Consider the display size, not just the pixel dimensions. A 2400×1600 photo displayed at 800×533 on a website can tolerate lower quality because the browser's downscaling smooths over minor artifacts. The same image used as a full-screen background or printed at A3 needs a higher quality setting. Match compression aggressiveness to how the image will actually be viewed.

Common mistakes that cause blur

1. Treating the quality slider as linear

JPEG quality is not linear. The difference between quality 95 and quality 90 is barely visible. The difference between quality 75 and quality 70 is dramatic. Most of the perceptible quality loss happens below 80%, and most of the file size savings also happen below 85%. This means you should almost never need to go below 75% for any image meant to be viewed — and never below 70% for anything other than tiny thumbnails. If you're going below 70% trying to hit a file size target, the problem is likely the image dimensions, not the quality setting.

2. Re-saving a JPEG that was already compressed

Opening a saved JPEG in Photoshop, making a small crop, and re-exporting at "high quality" still runs the compression algorithm again on top of existing artifacts. The result is a file that looks slightly worse than either the original or a fresh export. Every re-save of a JPEG — even at 100% quality — introduces a small amount of additional degradation. Always keep an uncompressed master (PSD, TIFF, or high-quality export) and only produce the compressed JPEG once, as the final step.

3. Compressing images that shouldn't be JPEGs

JPEG compression was designed for photographs — images with continuous color gradients, subtle tonal variations, and no hard edges or text. Apply JPEG compression to a screenshot with text, a logo, a diagram, or any image with large solid-color areas, and the artifacts are immediately visible: fuzzy text, color fringing around edges, mossy noise in flat fills. These images should be PNG or WebP. The rule is simple: photographs → JPEG or WebP. Everything else → PNG or WebP.

4. Ignoring the encoder

Most image editing software — including older versions of Photoshop — uses the standard libjpeg encoder. MozJPEG produces noticeably sharper results at the same file size, or the same visual quality at a smaller file size. If you're using an encoder that doesn't mention MozJPEG and you care about quality-to-size ratio, you're leaving significant efficiency on the table.

5. Not resizing before compressing

If you're compressing a 4000×3000 photo for a blog post that displays it at 1200×900, you're compressing far more pixels than will ever be shown. Resize the image to its actual display dimensions (or 2× for retina displays) before compressing. A correctly sized image at quality 82% will always look sharper and be smaller than an oversized image at any quality setting, because the browser doesn't have to perform its own potentially lossy downscaling at render time.

Real-world compression examples

These results use MozJPEG at the specified quality settings, with metadata stripped and progressive encoding enabled. Source files were unoptimized camera exports:

The pattern holds across image types: MozJPEG at 75–85% with metadata stripped consistently delivers 50–70% file size reduction with no visible quality difference at normal web viewing sizes. The exception is content with fine text or hard edges, which shouldn't be JPEG at all.

Tool comparison

The right tool depends on your workflow. Browser-based tools are ideal for one-off compression; CLI tools and libraries suit production pipelines where automation matters.

Compress your JPEG right now — free

The Rebrixe JPEG Compressor runs MozJPEG entirely in your browser using WebAssembly. Your images are never sent to a server — compression happens locally on your device. No account required, no file size limit, no watermarks. Set your target quality, preview the result side by side, and download when satisfied.

Frequently asked questions