Compress images to reduce file size while maintaining visual quality. Supports lossy and lossless compression with real-time preview and size comparison.
JPEG files on the open web average 100-300 KB, but a single uncompressed photo from a modern phone sensor can exceed 8 MB. That 20-40x gap is almost entirely compression headroom, and most images saved at quality 100 carry detail the human eye cannot resolve at normal viewing distances. This compressor lets you reclaim that waste interactively. You set a JPEG, PNG, or WebP quality target, see the result side by side with the original, and decide whether the tradeoff is worth it before committing to a download. The slider maps directly to the quantization matrix in JPEG's DCT pipeline. At quality 95, the matrix rounds frequency coefficients so gently that output is nearly indistinguishable from source. At quality 50, high-frequency coefficients collapse to zero, producing visible softness on fine textures and ringing artifacts around sharp edges. For most web use, 75-85 is the sweet spot where file size drops 40-60% and differences are hard to spot without zooming past 200%.
Initializing in your browser…
Optimize and minify SVG files by removing unnecessary metadata, comments, and whitespace while preserving visual quality.
Convert SVG vector graphics to high-quality raster images (PNG, JPG, WebP) with custom scaling and background options.
Resize and scale images with smart aspect ratio control. Supports custom dimensions, percentage scaling, social media presets, and batch resizing while maintaining image quality.
A blog hero is 3.2 MB and tanking your mobile load time; you need it small but still sharp.
Input
cover.jpg 3.2 MB · target quality 75
Output
cover.jpg 410 KB, ~87% lighter, no visible artefacts at this quality
Lowering JPEG quality discards detail the eye barely registers; 75 is usually the sweet spot before banding appears. A live before/after lets you push compression exactly as far as the image tolerates.
JPEG files on the open web average 100-300 KB, but a single uncompressed photo from a modern phone sensor can exceed 8 MB. That 20-40x gap is almost entirely compression headroom, and most images saved at quality 100 carry detail the human eye cannot resolve at normal viewing distances. This compressor lets you reclaim that waste interactively. You set a JPEG, PNG, or WebP quality target, see the result side by side with the original, and decide whether the tradeoff is worth it before committing to a download. The slider maps directly to the quantization matrix in JPEG's DCT pipeline. At quality 95, the matrix rounds frequency coefficients so gently that output is nearly indistinguishable from source. At quality 50, high-frequency coefficients collapse to zero, producing visible softness on fine textures and ringing artifacts around sharp edges. For most web use, 75-85 is the sweet spot where file size drops 40-60% and differences are hard to spot without zooming past 200%.
Compress hero images, product photos, and thumbnails to speed up page loads and improve Core Web Vitals scores.
Shrink photos to fit within email size limits without resorting to zip files or cloud links.
Reduce the space your photo library uses on disk or cloud storage.
Create lighter images that load fast on cellular connections.
JPEG compression has two main levers. The first is the quantization step applied to the output of an 8x8 discrete cosine transform on each block of pixels. Higher compression rounds more aggressive frequency coefficients to zero, which the entropy coder then stores in far fewer bits. The second is chroma subsampling: 4:2:0 subsampling cuts color resolution to a quarter of luma resolution, which the human visual system barely notices on photographic content but which halves the color data outright. This tool applies both in a coordinated way that matches libjpeg's defaults at each quality level.
PNG is a different beast. It is lossless, pixel data round-trips bit for bit, so "compression" here means optimizing the DEFLATE pipeline: choosing a better filter per scanline, rebuilding the Huffman tables, and optionally reducing the palette for images that do not actually need 16.7 million colors. For flat illustrations, screenshots, and line art, an optimized PNG can shrink 30-50% with zero visual change. Photographic content rarely compresses well as PNG regardless of optimization because DEFLATE cannot exploit the spatial redundancy that DCT handles so efficiently.
WebP sits between the two. In lossy mode it uses VP8's block prediction plus transform coding, typically producing files 25-35% smaller than equivalent-quality JPEGs. In lossless mode it uses LZ77-style matching plus color indexing, often beating optimized PNG by another 20-25%. The catch is slightly slower decode and compatibility with very old browsers. For modern web delivery, WebP at quality 80 is a strong default for most photographic content, and PNG remains the right choice for anything with hard edges, text, or transparency that chroma subsampling would destroy.
It depends on the image and format, but 50-80% reduction is typical for JPGs at quality settings that still look great. PNGs may see smaller reductions since they are already lossless.
JPG compression is always lossy. PNG optimization is lossless. WebP supports both modes. The quality slider controls the lossy trade-off.
For most web images, 75-85% quality is a good balance. Below 60%, artifacts start becoming noticeable. Use the preview to judge for your specific image.
Often yes, especially for JPGs saved at maximum quality. Reducing from 100% to 85% quality typically cuts file size in half with no perceptible difference.
Images are decoded, edited, and exported entirely inside this browser tab. No originals, exports, or metadata are uploaded.